Issuu

Vaccine Trust Survey report Methodology and Key Results Read more: www.thevaccinetrustproject.com

I N T R O D U C I N G

T H E

T R U S T

M O D E L

This work set out to build a tool to better understand, measure, and enable targeted trust-building efforts in a health-seeking context—critical for building demand for health services

“Trust is an area where governments can move the needle, and the fact that it outweighs traditional measures of healthcare capacity and pandemic preparedness should be a wakeup call for all of us” Thomas J. Bollyky, director of the CFR global health programme & author of a recent Lancet study showing a link between trust and COVID-19 vaccination rates

Situation: trust as a critical – but poorly understood demand driver Trust is increasingly in-focus as a critical component in efficiently health-seeking behavior – but it is also clear that there is no consensus on how to understand trust today

Ambition: a trust-building a survey tool for decision-makers The ambition of this work is to develop a tool capable of helping decision-makers (e.g. NGO’s, governments, funders, and donors) harness the power of trust-based health interventions – ultimately driving efficient vaccine uptake through increased demand 2

I N T R O D U C T I O N

O F

T H E

T R U S T

F R A M E W O R K

T RU S T I N T H E …

Recap: The ethnographic research led to a trust model with four types of trust can be understood in isolation and through their relationships with each other—we call these quadrants and the relationships between them The Vaccine Trust Framework

Health system promise

Vaccine promise

Does the health system have my and my community’s best interests at heart?

Do I believe that this vaccine has value for me and my community?

A U T O N O M Y F A I R N E S S P R I O R I T Y

In isolation, each quadrant can help explain vaccine uptake, and identify where to target trust-building efforts to ensure vaccine uptake…

… and combined, the four trust types form a holistic understanding of trust that can more robustly help predict and understand vaccine behavior

Zooming in one type of trust in isolation can provide deep understanding of the barriers to vaccine uptake and where to target trust efforts.

Looking at the four trust quadrants combined provides a robust understanding of trust that can help predict vaccine uptake – and shed light on the quadrants' interaction

B E N E F I T A L I G N M E N T

A C C E S S

R E L E VA N C E S A F E T Y

Healthcare delivery

Vaccine delivery

Does the health system generally work for me and my community?

Do I feel this vaccine is available and accessible to me and my community?

C A P A B I L I T Y A F F O R D A B I L I T Y C O M P E T E N C E

A D E Q U A C Y

O F

C O M P A S S I O N

D E L I V E RY

S E T T I N G

I N F O

C O N F I D E N T I A L I T Y

A G E N C Y

I N T R O D U C T I O N

O F

T H E

T R U S T

F R A M E W O R K

Each trust types consists of multiple trust dimensions Health system promise

Vaccine promise

The perception that the health system recognizes people’s autonomy in making decisions about their own health

AU TO N O M Y

The perceived value of the vaccine to child and community

B E N E F I T The perception that the health system provides services in a non-discriminatory manner

FA I R N E S S

P R I O R I T Y

A L I G N M E N T

The perception that the health system works towards the same kinds of health outcomes as people The health system’s perceived ability to deliver on people’s expectations of treatment of issues that fall within the purview of the system

C A PA B I L I T Y

The perceived relevance of being protected against disease for child and community

R E L E VA N C E

The perceived risk of side effects or other adverse events

S A F E T Y

Healthcare delivery AC C E S S

The perceived ease of accessing healthcare – incl. distance, time, navigation, language barriers, and availability of medical provisions

A F F O R DA B I L I T Y

The perceived ability to get healthcare when needed without having to forego or delay treatment due to cost

C O M P E T E N C E

The perception that healthcare providers have the knowledge and skills required to attend to people’s issues

C O M PA S S I O N

The perception that providers engage patients with respect and recognition, and demonstrate a commitment to their betterment

C O N F I D E N T I A L I T Y

The perception that medical and personal information will be kept private and undisclosed outside the provider/patient relationship

Vaccine delivery A D E QUAC Y

O F

I N F O

The perceived completeness of the information provided about the vaccine

D E L I V E RY

S E T T I N G

The percieved appropriateness of the site(s) where the vaccine is delivered, including medical competence and safety when accessing the vaccine

AG E N C Y

The percieved adequacy of consent collection, including whether people trust they will be asked for their consent and whether their decision will be respected

I N T R O D U C T I O N

O F

T H E

T R U S T

F R A M E W O R K

The Vaccine Trust Framework offers a picture of the trust landscape at a specific time—over time, it gives insights into the dynamic interactions among the four trust quadrants EXAMPLE:

MANDATED

YEAR 1 Health system promise

Healthcare delivery

VACCINES

YEAR 3

Vaccine promise

Vaccine delivery

A vaccine is pushed through despite low vaccine trust e.g., by mandate…

Health system promise

Healthcare delivery

YEAR 5

Vaccine promise

Vaccine delivery

… which decreases the trust in the health system over time…

Health system promise

Healthcare delivery

Vaccine promise

Vaccine delivery

… creating even lower trust levels for future vaccines

This is a key component of the Vaccine Trust Framework within low-trust contexts:

Every intervention has a trust impact – positive or negative – which eventually impacts trust in the overall health system 5

I N T R O D U C T I O N

R E P O R T

O V E R V I E W

This report outlines the methodological choices behind the Vaccine Trust Framework and presents the key findings on trust and vaccine uptake in Kenya and Pakistan Chapters Methodology

Sample composition

Quantifying the framework

Differences in trust

Trust and the effect on vaccine uptake

HPV learnings

METHODOLOGY

SAMPLE COMPOSISTION

QUANTIFYING THE FRAMEWORK

DIFFERENCES IN TRUST

HPV LEARNINGS

• Summary statistics on key demographic and socioeconomic variables

• Calculate and present trust scores for Kenya and Pakistan

• Explore differences and in trust across key demographic and socioeconomic variables

TRUST AND THE EFFECT ON VACCINE UPTAKE

•

Operationalization of the Trust Framework Sampling strategy

• Key statistics on health seeking behaviors incl. vaccine rates

• Internal and external validation of the Vaccine Trust Framework

• Present the effect of trust on vaccine uptake • Explore how different trust types interact with each other • Explore the effect of trust on vaccines across genders and marginalized groups

• Present key learnings om HPV vaccine perceptions and key influencers in Kenya • Present preliminary findings on HPV and HPV vaccine awareness in Pakistan

Methodology Sample composition Quantifying the Trust Framework Differences in trust Trust and the effect on vaccine uptake HPV learnings 7

C H A P T E R

O V E R V I E W

In the following chapter, we outline the methodological choices behind the Vaccine Trust Framework and the survey strategy S U M M A RY

CHAPTER OUTLINE

Challenges with existing trust measurements We outline challenges with existing trust measures and how the Vaccine Trust Framework seek to improve these challenges

Operationalization

We walk through the process of operationalizing the survey, crafting the questions and outline key definitions

Testing and validating the survey tool

Sampling strategy and data processing

We lay forward how the survey tool itself was tested and validated to ensure high data quality

We present the sampling strategy and why we report on non-weighted data

The operationalization of the Vaccine Trust Framework is based on an extensive literature review and two rounds of ethnographic research in Kenya and Pakistan. The survey was reviewed by experts and tested extensively before going to the field. The survey strategically oversamples caregivers to girls aged 10-14 across Kenya and Pakistan. The results are unweighted. 8

M E T H O D O L O G Y

T R U S T

L I T E R AT U R E

Recap | Our review of the trust literature revealed four major challenges with approaches to measuring trust in a health system and vaccine context CHALLENGES

Trust measures are situational and imprecise

There is a lack of agreement on what trust is

Most trust measures are Western-centric

Trust measures are removed from decision-making

Existing trust measures are developed in an ad hoc manner and are rarely grounded in a deep, contextual understanding of trust

There is disagreement within the quantitative trust literature on what trust is, which dimensions are important, and how to measure them

Existing trust measures are developed and validated in a Western context with insufficient attention to potential LMIC specificities

Existing trust measures devote little attention to solutions and decisionsmakers’ perspectives and are rarely applied outside of academia

“More work should also be done to improve existing trust measures. Validity of the measures could be strengthened by using qualitative methods and pilot-testing scales and indices.” 1

“People's trust in the health system plays a role in explaining one’s adherence, access to and utilization of medical care [...]. Yet it is not easy to find trust measures and understand what they are measuring.” 2

“While we found growing numbers of health systems trust measures, very few were developed and validated in lowand middle-income countries.” 3

“Improving well-being requires solid evidence that can inform policymakers and citizens where, when, and for whom life is getting better… Nevertheless, certain topics have not yet received the attention […]Trust is one of these topics.” 4

1) Ozawa, Sachiko, and Pooja Sripad. "How do you measure trust in the health system? A systematic review of the literature." Social science & medicine 91 (2013): 10-14. 2) *How do you measure trust in the health system? A systematic review of the literature. Ozawa & Sripad. Soc Sci Med. August 2013

3) Ozawa, Sachiko, and Pooja Sripad. "How do you measure trust in the health system? A systematic review of the literature." Social science & medicine 91 (2013): 10-14. 4) OECD Guidelines on Measuring Trust. OECD. 2017

M E T H O D O L O G Y

T R U S T

L I T E R AT U R E

Recap | The Vaccine Trust Framework survey seek to improve each of these to deliver a data tool that is optimized for decision makers in LMIC contexts CHALLENGES

Trust measures are situational and imprecise

There is a lack of agreement on what trust is

Most trust measures are Western-centric

Trust measures are removed from decision-making

SOLUTIONS • The survey has been developed based on two rounds of detailed qualitative research

• Health- and vaccine-related trust have been operationalized meticulously

• The survey has undergone both

to capture their multiple dimensions and optimize measurement validity

ensure that it’s fit-for-purpose and that central concepts are understood • The survey integrates trust and vaccine measures to bring the two fields together

• Existing validated items from the literature have been used

expert review, cognitive testing, and pilot testing to

to the extent possible to further advance the field

• The survey has first and foremost been developed and validated for use in LMIC contexts, which have particular dynamics that aren’t captured by Westerncentric measures • The survey enables testing of hypotheses linking trust to vaccine acceptance in LMIC

contexts – and identification of relevant proxies for situations where a particular vaccine hasn’t been introduced yet

• Key decision makers both globally and nationally have been engaged from the outset

of the work to design a relevant data tool for their use cases • Data has been collected to not

only map trust nationally but also regionally, with an aim to

inform targeted approaches and vaccines to drive trust and vaccine uptake

M E T H O D O L O G Y

O P E R AT I O N A L I Z AT I O N

Operationalization

Expansion

HEALTH SYSTEM

VACCINES

DIMENSION #1

SUB-DIMENSION #1  Items SUB-DIMENSION #...

HPV COVID-19

Combining the insights from the C19 and HPV work has revealed imperatives for measuring trust validly EXAMPLE

INSIGHT

Access to healthcare is more than just distance to a clinic – it’s about knowing where to go in the first place, understanding the language and not having to wait for hours

PROCESS

PROMISE

Through ethnographic fieldwork in Kenya and Pakistan, the Vaccine Trust Framework has been operationalized and expanded to an additional vaccine; the HPV vaccine

S U RV E Y I M P L I C A T I O N

Measure trust in various sub-dimensions of access Operationalization has entailed defining the dimensions of each quadrant in a data-driven way and specifying valid indicators to measures them

Expansion has entailed shifting the focus from C19 to HPV vaccination to understand its trust dynamics and qualify the Trust Framework

The questions on access includes distance to clinic, langugae barriers, wait time and overall difficulty of accessing healthcare and was validated during the cognitive testing and pilot test.

M E T H O D O L O G Y

O P E R AT I O N A L I Z AT I O N

Systematically working from the C19 & HPV research and literature in the field enabled identification of the dimensions that constitute trust in each quadrant of the framework C19

RE SE ARCH

HPV

RESE ARCH

LIT

REVIEW

T RU S T

DIMENSIONS

M E T H O D O L O G Y

O P E R AT I O N A L I Z AT I O N

The trust quadrants and dimensions proceeded by breaking dimensions into sub-dimensions and finally survey items to measure them From quadrants to dimensions and sub-dimensions…

… measured through sets of items

items to measure the quadrants Trust Measurement Package Documents literature review

Where possible, validated items from the literature were used to build on best practice, enable comparisons, and contribute to cumulative knowledge. However, because of the granularity of the trust framework, a range of new items also had to be created and tested.

+75

items capturing, e.g., demographics, family composition, and vaccine behaviors were also added to fulfil the secondary aims 13

M E T H O D O L O G Y

O P E R AT I O N A L I Z AT I O N

Beyond trust items, a secondary aim was to gauge opinions on the HPV vaccine – and the questions was developed based on qualitative findings and existing HPV survey items EXAMPLE

HPV

QUESTIONS

Examples of HPV specific questions

Please tell me if you do not at all agree, somewhat agree, or strongly agree: “Vaccines that are specifically made for adolescent girls are suspicious”

Please tell me if you do not at all agree, somewhat agree, or strongly agree: “I am concerned about serious side effects of the HPV vaccine” Please tell if you “Would you accept the HPV vaccine if it was recommended by… Your family? Your friends? Their teacher? Your church? A healthcare provider? Please tell me why you did not accept the HPV vaccine: Examples of options: “The HPV vaccine is not effective”, ”My daughter is not at risk of getting cervical cancer”, “I am against vaccines in general”

Please tell me why you did accept the HPV vaccine: Examples of options: “The HPV vaccine will protect my daughter's future”, “The HPV vaccine prevents my daughter from getting HPV”,

46 items related to HPV

By combining learnings from the qualitative research and existing HPV survey items we ensured that our questions captured the most important HPV dynamics… QUA L R E S E A RC H

E X I S T I N G S U RV E Y S

Our research revealed existing mental models around the HPV vaccine, cervical cancer and barriers to uptake. These formed the foundation for which questions to ask and the answer categories e.g., on why/why not people had accepted the vaccine. When possible, we used participants own words and descriptions to ensure the questions were understandable to the survey audience.

We reviewed existing HPV surveys and identified common categories of questions incl. fear of side effects, trust in the HPV and trust in the benefits of the vaccine to ensure we didn’t miss important categories of questions. Most HPV surveys are developed in a European or North-American context so, we did not carry any questions directly over to the Trust Survey but changed language following a no harm principle, to ensure our survey didn’t raise participants’ concerns about the HPV vaccine. 14

M E T H O D O L O G Y

O P E R AT I O N A L I Z AT I O N

Rather than structuring the survey strictly according to the logic of the Vaccine Trust Framework, it has been structured to have a meaningful flow for respondents The survey consists of 11 main sections SCREENING

HPV AWARENESS & VACCINATION STATUS

PERCEPTION OF THE HEALTHCARE SYSTEM

ACCESS TO HEALTHCARE SERVICES

PERCEPTION OF HEALTHCARE PROVIDERS

ATTITUDES TOWARDS HPV VACCINES

ATTITUDES TOWARDS C19 VACCINES

HEALTHSEEKING BEHAVIOR

BACKGROUND QUESTIONS AND HOUSEHOLD INFO

GENERAL TRUST QUESTIONS

157

items in total

Singlechoice

Multiplechoice

QUESTION TYPES

REDUCED FULL

Likert scales

ATTITUDES TOWARDS CHILDHOOD VACCINES

Designed for a CAPI mode of administration – taking ~45 minutes to complete

A sub-aim of this initial proof-of-concept run of the survey is to identify how best to compress it to around ~20 minutes to also be suitable for other modes of administration. Individual items/sections can also be included in other surveys.

M E T H O D O L O G Y

O P E R AT I O N A L I Z AT I O N

Key socio-demographic variables are defined based on best practice and with input from experts…

KEY DEFINITION

VA L I DAT E D S O U RC E

CAREGIVERS Adults who are responsible for the well-being of children, which could include: making financial, educational, or health-related decisions on behalf of children, being engaged in the raising, rearing, or bringing up of children, or responsible for disciplining children. The caregiver does not necessarily have to be the parent of a child.

Based on recommendation from the Ipsos team incl. LMIC survey experts

WORK Engaging in activities for which you are paid in cash or kind. The question is followed up with the question “What work do you do?” with country specific answer categories developed with input from the project’s Gender Advisors.

Based on the World Bank’s definition of work1 and adjusted based on input from the project’s Gender Advisors 1) https://databank.worldbank.org/metadataglossary/worlddevelopment-indicators/series/SL.IND.EMPL.ZS

VULNERABILITY (KE) Vulnerability in Kenya is defined according to the Pathway Vulnerability Typing Tool.

Borrowed from the Pathway Typing Tool 16

M E T H O D O L O G Y

T E S T

VA L I D AT I O N

… and different experts reviewed and provide feedback on the survey from their respective vantage points to Institutional Review Boards Consultative group composed of experts from modeling, immunization, GE, and PHC Suggested including vulnerability items from the PATHWAYS typing tool

Advisors with both subject matter, survey design, and sampling expertise Suggested a stratified multi-stage clustered sample design to enable both national and regional analyses

The survey was reviewed by an IRB in both countries and approved by relevant authorities Suggested minor language tweaks and ensure ethical standard of the survey CROSS-INSTITUTIONAL PANEL

Gender Advisors Two gender experts – specialized in Prof. Emma Varley Rhoda Maina Pakistan and Kenya respectively – have been advising the work all throughout and applied their deep contextual understanding to the survey.

Authorities in the field of vaccine confidence and HPV vaccination from the Vaccine Confidence Project incl. Professor Heidi Larrson reviewed the survey with particular attention to comparability with existing measures.

Experts from a community of interest around measuring trust in a health context – and sociobehavioral drivers more generally – provided feedback based on their own and their institutions’ experience with measuring complex social phenomena.

EXAMPLE CHANGE

A set of general vaccine confidence / hesitancy questions from VCP’s work were included

Sharpen question formulation and identify key trust measures to include.

Include additional categories of employment to capture different women’s experiences.

M E T H O D O L O G Y

T E S T

VA L I D AT I O N

The survey went through multiple rounds of testing to optimize it in terms of relevance, comprehensibility, and feasibility GENERATIVE

TESTING

• During ethnographic fieldwork • Building the survey on contextual understanding

COGNITIVE

TESTING

• 16 in-depth cognitive tests carried out • Focusing on expected challenging parts of the survey

PILOT

TESTING

• 289 tests carried out (KE = 192; PK=97) • Testing the full survey to simulate the final experience

PURPOSE

Getting an initial sense of people’s ability to answer trust and vaccine-related questions by testing an alphaversion of the survey that was developed based on hypotheses from the C19 work and other existing survey instruments.

Testing the comprehensibility of specific parts of the survey to assess the reliability of responses and identify issues. An example was whether respondents understood and answered based on the definition provided of “abstract” concepts such as ‘health system’.

Simulating the experience of answering the survey in its final form with a larger group of pilot respondents to identify parts for trimming, test phrasing and translations, evaluate the structure/flow/filters, and refine response categories.

EXAMPLE IMPLICATIONS

• Making informed decisions about whether to refer to ‘system’, ‘authorities’, or ‘providers’ • Avoiding perceived insensitive HPV questions

• It was verified that respondents generally understood and could answer in relation to the ‘health system’ and distinguish between ‘community’ and ‘hospital’ providers • Particularly in Kenya, a 5-pt response scale posed difficulties in Swahili – therefore, a 3-pt scale was used instead for most items

• Compression of a few trust sub-dimensions – e.g., skills & education weren’t sufficiently distinct to warrant separate items • Removal of a few response categories in multiplechoice matrices to make them more manageable – e.g., when asking who people trust 18

M E T H O D O L O G Y

S A M P L I N G

S T R AT E G Y

D ATA

P R O C E S S I N G

The sampling strategy was designed to produce a nationally representative sample of caregivers of children aged 10-14, plus representative samples of select areas of interest The primary goal of the study is to achieve a nationally representative sample for caregivers of children age 1014 in Kenya and Pakistan* To obtain national representation, we employed stratified multistage cluster sampling in collaboration with Ipsos Global Stratification & sample sizing: We first stratified Kenya and Pakistan into 13 and 9 strata respectively. Sample sizes prioritized six strata of interest in Kenya, and five in Pakistan. Remaining strata were allocated sample sizes proportionate to their populations with slight reductions due to prioritized areas. Selecting sampling units: We divided each stratum into primary sampling units, then secondary sampling units, and finally, blocks as the lowest cluster level Selection via random walks: At the final cluster level, we screened every kth household for adolescent children, collecting demographic information for each household. For each cluster, we conducted 8-9 interviews with qualifying households *) For security reasons, we excluded military areas in Pakistan and some areas in the Northeastern part of Kenya

The secondary goal is to produce representative samples of geographical areas of interest

PA K I S TA N

Lahore Karachi Islamabad

Azad Jammu and Kashmir Gilgit-Baltistan Balochistan

K E N YA

Nairobi Kisumu Kilifi

Kitui Isiolo Uasin Gishu

M E T H O D O L O G Y

S A M P L I N G

S T R AT E G Y

D ATA

P R O C E S S I N G

The definition of caregiving was kept deliberately inclusive to accommodate diverse family structures, with caregivers responding on behalf of a child in the household

We assumed a broad definition of caregiving to ensure the representation of caregivers beyond parents1

DEFINITION A caregiver is an adult who either: • • •

Makes financial, educational, or healthrelated decisions on behalf of children Engages in raising, rearing, or bringing up children Is responsible for disciplining children

ELIGIBLE

Each caregiver was asked to answer on behalf of either a boy or a girl 1) See the exact definition on the next slide 2) This practice was adopted once fieldwork had begun; first field guide version employed randomized selection with 70% chance of selection allocated to girls

SELECTED

If both a boy and a girl between the ages of 10-14 were part of the household, caregivers were asked to answer on behalf of the girl2 20

M E T H O D O L O G Y

S A M P L I N G

S T R AT E G Y

D ATA

P R O C E S S I N G

As a result, the Vaccine Trust Survey offers insight into an understudied yet crucial population— caregivers to adolescents—who are key to understand vaccine behavior W H AT W E H AV E D O N E …

W H AT O T H E R S H AV E D O N E …

Populations  

Male caregivers to adolescents Female caregivers to adolescents

Vaccine behaviors     

COVID-19 vaccine status HPV vaccine status Childhood vaccine status Zero dose status HPV vaccine awareness

Trust patterns      

Trust in the health system promise Trust in the healthcare delivery Trust in vaccine promises Trust in vaccine delivery Interpersonal trust Institutional trust

FinAccess Household survey

 Female caregiver  Male caregivers × No trust data × No vaccine data Kenya 2021 & Pakistan 2018 (Wave 7)

Kenya 2021

The Social & Living Standards Measurement Survey Pakistan 2018

Kenya 2021 & Pakistan 2018

× Only childhood vaccine status children under 5yo × No HPV vaccine status × No HPV awareness × No trust data

× Only childhood vaccine status children under 3yr × No link between household data & vaccine data × No trust data

Global Monitor: Kenya 2018 & Pakistan 2018

Kenya 2022 (Round 8)

 Institutional trust & interpersonal trust  Trust in healthcare system × No vaccine data

 Institutional trust & interpersonal trust  Trust in healthcare system × No caregiver data × No vaccine data

 Institutional trust & interpersonal trust  Trust in healthcare system × No caregiver data × No vaccine data 21

M E T H O D O L O G Y

S A M P L I N G

S T R AT E G Y

D ATA

P R O C E S S I N G

All the results are calculated based on unweighted data to ensure transparency – and comparing weighted and unweighted results does not revealed significant differences C L E A N I N G DATA

W E I G T I N G T H E DATA

The final data set has been cleaned including the following steps:

This report presents the unweighted data to comply with best practices and ensure transparency around our results

• The data has been anonymized • Outliers have been removed based on income in Kenya • Variables have been recoded to have the same direction

There is very little difference in the trust scores in both Kenya and Pakistan when comparing the weighted and unweighted* results – suggesting a limited benefit of weighting the data PAK I STAN

”Allowing readers to understand who was interviewed and who was excluded, and to contextualize results, accordingly, may often be far more informative than any statistical adjustment done "behind the scenes".” - Collins et al. 20221

Unweighted

K E N YA

Weighted Unweighted

Weighted

Health system promise

Healthcare delivery

Vaccine promise

Vaccine delivery

*Weights are calculated based on income. The weights are available upon requests.

1) Collin et al. 2022 “Representativeness of remote methods in LMICs: A cross-national analysis of pandemic-era studies

Methodology

Sample composition Quantifying the Trust Framework Differences in trust Trust and the effect on vaccine uptake HPV learnings

C H A P T E R

O V E R V I E W

In the following chapter, we present the sample on key socio-demographic variables and vaccine rates and outline key considerations about the sample S U M M A RY

CHAPTER OUTLINE

Sociodemographic variables

Health-seeking behavior and vaccine rates

We present the sample distribution on key sociodemographic variables incl. gender, age and income

We present key statistics on health seeking-behavior incl. vaccine rates on childhood vaccines, COVID19 vaccine, and the HPV vaccine in Kenya.

Geographical spread

Sample issues and key considerations

We outline the geographical spread of the sample on a regional level

Lastly, we highlight key challenges with the sample distribution and key considerations

•

The sample consists of primarily younger female caregivers to adolescents aged 10-14

•

In both Kenya and Pakistan, the sample skews towards higher vaccination rates

•

In Kenya, the sample skews slightly toward higher-income groups

S A M P L E

C O M P O S I T I O N

S U M M A RY

S TAT I S T I C S

K E N YA

Demographic & socioeconomic profile: Randomized sampling of caregivers to adolescents has resulted in a majority of younger, female respondents K E N YA n = 3 6 7 0

Age

Income

Median 36 years

Median 10 000 KES/month

42%

PETTY TRADER/SELF-EMPLOYED

26% 16% 9%

PROFESSIONAL, TECHNICAL, MANAGERIAL HOUSEHOLD, DOMESTIC, AND SERVICES

CLERICAL OR SALES

OTHER

UNEMPLOYED/RETIRED/HOUSEWIFE

49%

>65

PROTESTANT

26%

Complete secondary

24%

CATHOLIC

14%

OTHER CHRISTIAN

13%

Incomplete secondary

11%

MUSLIM

22%

Complete primary

17%

Incomplete primary

No education

*) Past Ipsos studies give reason to believe that this split – based on random sampling of caregivers in households – is a representative reflection of the true distribution.

Religion 16%

Higher than secondary

>25

12%

Education

Type of work AGRICULTURAL

5% 45-54

BOYS

35-44

GIRLS

25-34

44%

<25

56%

SKILLED AND UNSKILLED MANUAL LABOR

55-64

Child gender

15-25

14%

10-15

31%

22%

17%

8-10

MALE

22%

Thousand KES

6-8

FEMALE

43%

4-6

20%

2-4

80%

0-2

Respondent gender*

NONE

TRADITIONAL AFRICAN

OTHER

S A M P L E

C O M P O S I T I O N

S U M M A RY

S TAT I S T I C S

K E N YA

Health & vaccine profile: More than half of the sample have vaccinated their daughters against HPV, while the vast majority accept childhood vaccines K E N YA n = 3 6 7 0

Public/private healthcare

COVID-19 vaccination

82%

18%

PUBLIC

PRIVATE

Type of health service used for most of respondent’s health issues

38%

YES

Vaccination status of respondent

76%

24%

YES

Whether respondent has heard about the HPV vaccine

Child vaccination

Type of provider

HPV vaccination

47%

LOCAL HOSPITALS

42%

HEALTH DISPENSARIES

27%

REFERRAL HOSPITALS

23%

PHARMACIST

DOCTORS WITH THEIR OWN OFFICE

62%

HPV vaccine awareness

12%

98% YES Whether child has received at least one dose of measles, hepatitis B, yellow fewer, or tuberculosis vaccine

2% NO

63%

37%

YES

Whether child is vaccinated against HPV

Type of healthcare provider used for most of respondent’s health issues (multiple choice) – top 5 across respondents 26

S A M P L E

C O M P O S I T I O N

S U M M A RY

S TAT I S T I C S

PA K I S TA N

Demographic & socioeconomic profile: Randomized sampling of caregivers to adolescents has resulted in a majority of female respondents across income groups PA K I S TA N n = 3 7 3 4

Age Median 38 years

Respondent gender*

Type of work 30% 13%

SALARIED EMPLOYEE - GOV

11%

AGRICULTURAL

28%

Complete secondary

19%

Incomplete secondary

SKILLED LABOR CULTIVATOR

OTHER

15%

Incomplete primary

UNSKILLED LABOR

No education

*) Past Ipsos studies give reason to believe that this split – based on random sampling of caregivers in households – is a representative reflection of the true distribution.

20%

>150

SHIA OTHER MUSLIM CHRISTIAN

Complete primary

84%

SUNNI

11%

Higher than secondary

Religion

21%

SALARIED EMPLOYEE - PRIVATE

100-150

80-100

30-50

15-30

45-54

35-44

12% 6%

Education

PETTY TRADER/SELF-EMPLOYED

UNEMPLOYED/RETIRED/HOUSEWIFE

25-34

BOYS

2% <25

GIRLS

20%

50-80

23%

43%

32% 27%

Child gender 57%

Thousand PKR

51%

10-15

MALE

<10

FEMALE

>65

34%

55-64

66%

Income Median 30-50 Thousand PKR/month

5% 2%

HINDU

OTHER

NONE

S A M P L E

C O M P O S I T I O N

S U M M A RY

S TAT I S T I C S

PA K I S TA N

Health & vaccine profile: The majority of the sample is vaccinated against COVID-19 and accepts childhood vaccines for their children – while the HPV awareness is low, as expected PA K I S TA N n = 3 7 3 4

COVID-19 vaccination

Public/private healthcare

64%

36%

PUBLIC

PRIVATE

Type of health service used for most of respondent’s health issues

16%

YES

95% YES

Whether respondent has heard about the HPV vaccine

Child vaccination

HPV vaccination

57%

DOCTORS WITH THEIR OWN OFFICE

44%

REFERRAL HOSPITALS

38%

LOCAL HOSPITALS

LADY HEALTH WORKER

84%

Vaccination status of respondent

Type of provider

LOCAL HEALTH CLINIC

HPV vaccine awareness

17% 15%

91%

YES

N/A

Whether child has received at least one dose of measles, hepatitis B, yellow fewer, or tuberculosis vaccine

Type of healthcare provider used for most of respondent’s health issues (multiple choice) – top 5 across respondents 28

S A M P L E

C O M P O S I T I O N

S U M M A RY

S TAT I S T I C S

Geographically, the sample ensures representation across Kenya and Pakistan, including hard-toreach populations in the North Eastern Kenyan region and Balochistan in Pakistan PA K I S TA N

K E N YA

G I L G I T

R I F T

VA L L E Y

E A S T E R N N O R T H

298

461 514

156

K B Y B E R

688

E A S T E R N

W E S T E R N

B A L T I S T A N

869

P A K H T U N K W H A

119

A Z A D &

417

J A M M U

K A S H M I R

153

C E N T R A L N YA N Z A

369 352

P U N J A B

1697

N A I R O B I

B A L O C H I S T A N

C O A S T

S I N D H

178

978

S A M P L E

C O M P O S I T I O N

S U M M A RY

S TAT I S T I C S

Sample issue - higher income in Kenya: Our randomized sample of caregivers of adolescents skews towards higher vaccination rates and higher incomes compared to general population benchmarks K E N YA

Income 10 000 KES/month 22%

Thousand KES 12%

• 26% of sample below national poverty lines: Despite the apparent skew toward higher income, our sample does represent low-income caregivers using the Kenya Bureau of Statistic’s poverty lines for urban and rural areas2

>25

5% 15-25

6-8

4-6

8-10

2-4

0-2

22%

17%

10-15

Median

We have reason to believe that this relative skew toward higher incomes represents our population of caregivers

*) See appendix for sampling strategy + full sample composition

Income In 2021, the median income for parents (of children of all ages) was 5.000 KES/month (FinAccess) 1

1) FinAccess, 2021: “FinAccess 2021”. Median income calculated based on full dataset, see https://finaccess.knbs.or.ke/reports-and-datasets

• Our sample ended-up oversamples urban, and in turn, richer populations. Our sample design oversamples urban populations who generally have higher incomes.3 • Our sample population is likely to be professionally active. Parents to adolescents are often in the most productive age (mid twenties to late forties) (see appendix), and, therefore, more likely to be part of the workforce than other population groups. • The most recent benchmark contains data from mid-C19-pandemic when poverty in Kenya peaked, suggesting that the benchmark available may provide an underestimate of income levels for 20234

2) Kenya Bureau of Statistics (2021) defines poverty line as Ksh 3,947 and Ksh 7,193 per person per month for rural and urban areas respectively in World Bank 2023: Kenya Poverty and Equity Assessment 2023, https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099121323073037589/p1773530a7eb3009308e3f08663aa95c826 3) World Bank 2023: Kenya Poverty and Equity Assessment 2023, https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099121323073037589/p1773530a7eb3009308e3f08663aa95c826 4) World Bank 2023: Kenya Poverty and Equity Assessment 2023, https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099121323073037589/p1773530a7eb3009308e3f08663aa95c826 30

I N T R O D U C T I O N

T RU S T

DATA

Sample issue - higher vaccine coverage: In both Kenya and Pakistan, our randomized sample of caregivers of adolescents skews towards higher vaccination rates K E N YA

PAK I STAN

Vaccination rates

COVID-19 vaccination (of respondent)

62%

38%

84%

16%

YES

Childhood vaccination (of child)

98%

YES

91%

YES

EXTERNAL DATA

HPV vaccination (of child) 63%

37%

YES

HPV vaccine rate

COVID-19 vaccine

Childhood vaccine

COVID-19 vaccine rate

33% have received the first dose and 16% have received the second by 20221.

Childhood vaccine rate

Approx. 32% of adults had received two doses by 20222.

Approx. 68%3 of children are vaccinated.

Approx. 64% of the population has received two doses.4

The national vaccine rate is approx. 76% 5

1) Karanja-Chenge, Christine Muthoni, 2022: “HPV Vaccination in Kenya: The Challenges Faced and Strategies to Increase Uptake. Front Public Health. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8978582/ 2) Statista, 2022: “Share of adult population fully vaccinated against COVID19 in Kenya as of July 9 2022, by county”. https://www.statista.com/statistics/1252641/share-of-population-fully-vaccinated-against-covid-19-in-kenyan-counties/ 3) Allen et al., 2023: Inequalities in childhood immunization coverage associated with socioeconomic, geographic, maternal, child and place of birth characteristics in Kenya. BMC Infectious Diseases . https://bmcinfectdis.biomedcentral.com/articles/10.1186/s12879-021-06271-9/tables/

We have reason to believe that this relative skew toward higher vaccination rates reflects our population of caregivers of adolescents • Our sample population is likely to be one of the most active groups of society. Parents to adolescents are often in the most productive age (mid twenties to late forties), and so, some of the most active citizens • Our sample oversamples urban populations in both Kenya and Pakistan—a subpopulation that historically has been easier to reach in vaccine interventions. • Most recent benchmarks contain data from 2022. With data collection taking place in late 2023, vaccination rates for especially C19 and HPV may have improved since most recent benchmarks

4) Reuters COVID-19 Tracker, 2022: “Pakistan”. https://www.reuters.com/graphics/world-coronavirus-tracker-and-maps/countries-and-territories/pakistan/ 5) Unicef, 2021: “Country Office Annual Report 2021”. https://www.unicef.org/media/116341/file/Pakistan-2021-COAR.pdf

Methodology Sample composition

Quantifying the Trust Framework Differences in trust Trust and the effect on vaccine uptake HPV learnings

C H A P T E R

O V E R V I E W

In the following chapter, we build and test the validity of the Vaccine Trust Framework and its constitutive parts S U M M A RY

CHAPTER OUTLINE

Approach to building the Vaccine Trust Framework

Kenya and Pakistan’s Vaccine Trust Scores

We demonstrate how the Trust Framework is quantified bottomup, from questions to trust components.

We present the national trust scores for Kenya and Pakistan.

Imputing HPV vaccine trust in Pakistan

Testing the validity of the Vaccine Trust Framework

We impute HPV vaccine trust in Pakistan by using C19 and childhood vaccines as proxies.

We test the internal and external validity of the Vaccine Trust Framework to ensure its statistical existence.

The Vaccine Trust Framework is quantified bottom-up, and the imputed HPV vaccine trust score is calculated for Pakistan by averaging trust in childhood and the C19 vaccine. The validation of the Trust Framework confirms that the Trust Framework exists statistically speaking, and that its major components, levels and quadrants, for the most part strike a balance between consistency (correlation) and distinctiveness (holding unique variation in explaining trust as a phenomenon).

Q U A N T I F Y I N G

T H E

F R A M E W O R K

A P P R O A C H

The Vaccine Trust Framework assumes that each individual has their own trust profile and scores – aggregating the individual trust scores yields the trust score T RU S T S C O R E S RESPONDENT 1

T RU S T S C O R E S RESPONDENT 2

T RU S T S C O R E S RESPONDENT 3

K E N YA’ S

T RU S T

SCOR

69 67

The trust score is reported on a scale from 0 – 100, with 0 indicating low trust, and 100 indicating high trust*. * Statistically the trust score is normalized to a scale on 0 – 1, before being multiplying with 100 34

Q U A N T I F Y I N G

T H E

F R A M E W O R K

A P P R O A C H

To calculate the trust score, the Vaccine Trust Framework is quantified bottom-up progressing from dimensions, quadrants, levels, and ultimately, the full framework

From indicators to dimensions

From dimensions to quadrants

From quadrants to levels

Each dimension consists of a number of indicators, i.e., questions included in the survey

The Framework’s trust quadrants aggregate the scores of the trust dimensions for the quadrant in question

The level scores for the health system and vaccines respectively aggregate the quadrants that connect vertically

The full Trust Framework

Q U A N T I F Y I N G

T H E

F R A M E W O R K

A P P R O A C H

Out of the 150 questions included in the survey, 50 questions serve as indicators for the trust scores that aggregate into a trust score at the dimension level S U RV E Y Q U E S T I O N S

Please tell me if you do not at all agree, somewhat agree, or strongly agree: “The healthcare system prioritizes my good health” Please tell me if you do not at all agree, somewhat agree, or strongly agree: “The healthcare system prioritizes the health issues that are most important for me and my community”

PRIORITY ALIGNMENT

0 Low trust

100 High trust

The trust score is reported on a scale from 0 – 100, with 0 indicating low trust, and 100 indicating high trust*.

By averaging the values, we obtain a unified index for each trust dimension

Each dimension consists of 1-9 indicators * Statistically the trust score is normalized to a scale on 0 – 1, before being multiplying with 100 36

Q U A N T I F Y I N G

T H E

F R A M E W O R K

A P P R O A C H

The scores on dimensions are aggregated into the trust score on a quadrant level, again by averaging the individual’s trust score

PRIORITY ALIGNMENT

T RU S T S C O R E AT T H E Q UA D R A N T L E V E L

By calculating the average of the four dimensions that make up the quadrant for each individual and aggregating them, we obtain the trust score at the quadrant level —in this case, for trust in the health system promise

AUTONOMY

CAPA BI LI T Y FAIRNESS

Q U A N T I F Y I N G

T H E

F R A M E W O R K

A P P R O A C H

Averaging the two quadrant scores for the health system or the vaccine produces the trust score at either the health system or vaccine level 77 49

64 T RU S T

SCORE

77 H E A LT H

S YS T E M

H E A LT H C A R E

60 T RU S T

P RO M I S E

D E L I V E RY

T RU S T S C O R E AT T H E H E A LT H S YS T E M L E V E L

71 SCORE

47 56

In this case, we have calculated the trust score at a health system level by averaging each individual’s trust scores at the quadrant level

Q U A N T I F Y I N G

T H E

F R A M E W O R K

A P P R O A C H

92 By combining the two levels, we obtain the full Vaccine Trust Framework and the trust scores at the vaccine and system 77level

64 QUA DR ANT LEVEL

QUA DR ANT LEVEL

S YS T E M

H E A LT H C A R E

P RO M I S E

VA C C I N E

D E L I V E RY

VA C C I N E

QUADRANT SCORE

77 H E A LT H

HEALTH SYSTEM LEVEL

70 73

47 56

P RO M I S E D E L I V E RY

QUA DR ANT LEVEL

VA C C I N E L E V E L 39

Q U A N T I F Y I N G

T H E

F R A M E W O R K

T R U S T

S C O R E S

K E N YA

gn ali

omy

es riti

Be nef it

io Pr

n Auto

This approach produces the following framework and trust scores for Kenya, with the vaccine level for the HPV vaccine n me t

Ca pa

70 S Y S T E M

P RO M I S E

D E L I V E RY lity dentia Confi

on De liv er y

bility

Afford a

HEALTH SYSTEM LEVEL

ssi

Ac ce

a mp Co

se t

tin

VA C C I N E

P RO M I S E

VA C C I N E

D E L I V E RY

Adeq u

cy en Ag

en ce

e Saf

pe t

H E A LT H C A R E

nes

e nc

bil ity

Co m

H E A LT H

Fair

va le Re

acy o f info

VA C C I N E L E V E L 40

Q U A N T I F Y I N G

T H E

F R A M E W O R K

T R U S T

S C O R E S

K E N YA

n Auto

io Pr gn ali

omy

es riti

Be nef it

…And a total trust score of 69, compounding the health system and vaccine level

n me t

S Y S T E M

lity dentia Confi

HEALTH SYSTEM LEVEL

ssi

e Saf

eli ve ry

se t

tin

VA C C I N E

P RO M I S E

VA C C I N E

D E L I V E RY

Adeq u

G E N E R A L T R U S T S C O R ED

Ac ce

a mp Co

cy en Ag

D E L I V E RY

en ce

P RO M I S E

pe t

H E A LT H C A R E

Co m

H E A LT H

nes

bility

Fair

e nc

bil ity

Afford a

Ca pa

va le Re

acy o f info

VA C C I N E L E V E L 41

Q U A N T I F Y I N G

T H E

F R A M E W O R K

T R U S T

S C O R E S

PA K I S TA N

gn ali

o my

es riti

Be nef it

io Pr

n Auto

In Pakistan, the HPV vaccine remains to be rolled out—which, at first, leaves us with the trust score at the health system level alone n me t

ce an ev l Re

Ca pa

64 S Y S T E M

P RO M I S E

D E L I V E RY lity dentia Confi

De liv er y

se t

tin g

ability Afford

ncy

HEALTH SYSTEM LEVEL

acy o f info

i on ss

Ac ces s

a mp Co

Adeq u

e Ag

e Saf

H E A LT H C A R E

nes

Co m pe t

H E A LT H

Fair

bi l ity

Q U A N T I F Y I N G

T H E

F R A M E W O R K

I M P U T I N G

H P V

R E S P O N S E

However, we have observed that the HPV vaccine occupies a space between childhood and adult vaccines in fieldwork—and we see a similar pattern in the Kenyan survey data During fieldwork, we observed how the HPV vaccine occupied its own space in respondents’ mental models around vaccines…

This informed the hypothesis that trust in the HPV vaccine is neither entirely correlated with C19, nor childhood vaccines 1 – trust in HPV vac.

1 - trust in HPV vac.

1 – Trust in childhood vac.

βC19 0.24 βchildhood 0.68

Fieldwork in this project’s earlier phase suggested that the HPV vaccine targeting adolescent girls carries potential adult connotations while still being a vaccination decision made by caregivers on behalf of a child

βC19 0.55

0 Trust in Childhood vac. - 1

P-value <0.0001

Trust in C19 vac. - 1

P-value <0.0001

Both childhood and C19 vaccine items help explain unique variation in the response to the HPV vaccine in Kenya and remain significant when included in the same model —supporting the interpretation that both vaccine items can be used to impute HPV responses 43

Q U A N T I F Y I N G

T H E

F R A M E W O R K

I M P U T I N G

H P V

R E S P O N S E

With this in mind, we can impute the response to the HPV vaccine by averaging trust levels in COVID-19 and childhood vaccines B E N E F I T

T E S T I N G M E T H O D I N K E N YA

R E L E VA N C E

When averaging trust in childhood and C19 vaccines in Kenya to impute trust in the HPV vaccine, we find that each imputed dimension varies by a maximum of +/- 10 in trust score compared to observed scores

S A F E T Y

C AV E AT S

Pulling the average assumes that the context observed in Kenya—HPV occupying its own space due to its target group being adolescents—is also applicable to Pakistan.

FINDING

A D E QUAC Y O F I N F O

D E L I V E RY S E T T I N G S AG E N C Y

Calculating the average trust in childhood and C19 vaccines provides a relatively accurate picture of the observed HPV response in Kenya

Observed HPV response

An estimate based on childhood vaccines and C19 unifies two types of vaccines with diverse timelines, presenting a challenge in pinpointing when the imputed trust levels for the HPV vaccine would materialize. In particular trust in childhood vaccines appears to contribute to a slight overestimation of trust in the HPV vaccine.

Imputed HPV response 44

Q U A N T I F Y I N G

T H E

F R A M E W O R K

T R U S T

S C O R E S

PA K I S TA N

fit Be ne

en nm lig sa

o my

itie ior Pr

n Auto

This method provides us with the following framework and trust scores for Pakistan, including imputed trust scores for the vaccine level t

ce an ev l Re

Ca pa

64 S Y S T E M

lity dentia Confi

De liv er y

bility Afford a

HEALTH SYSTEM LEVEL

n sio

as mp Co

se t

tin g

VA C C I N E

P RO M I S E

VA C C I N E

D E L I V E RY

Adeq uacy of

cy en Ag

D E L I V E RY

Ac ce

P RO M I S E

nc e

H E A LT H C A R E

e Saf

Fair nes s

Co m pe te

H E A LT H

bil ity

info

VA C C I N E L E V E L 45

Q U A N T I F Y I N G

T H E

F R A M E W O R K

T R U S T

S C O R E S

PA K I S TA N

o my

en nm lig sa

Be ne

n Auto

itie ior Pr

fit

…And a compound trust score of 66 for Pakistan

ce an ev l Re

S Y S T E M

HEALTH SYSTEM LEVEL

n sio

e Saf

el ive ry

se t

tin g

VA C C I N E

P RO M I S E

VA C C I N E

D E L I V E RY

Adeq uacy of

G E N E R A L T R U S T S C O R ED ss

as mp Co

cy en Ag

lity dentia Confi

Ac ce

D E L I V E RY

nc e

H E A LT H C A R E

P RO M I S E

Co m pe te

H E A LT H

Fair nes s

bility

bil ity

Afford a

Ca pa

info

VA C C I N E L E V E L 46

Q U A N T I F Y I N G

T H E

F R A M E W O R K

I N T E R N A L

VA L I D AT I O N

The Trust Framework’s components are interrelated, but also all uniquely necessary to conceptualize trust—striking a balance between consistency and distinctiveness

FUL LY DI STI NC T

F U L LY C O N S I S T E N T

Producing fully consistent quadrants and dimensions, i.e., obtaining perfect correlation, would indicate that the Framework’s different components are obsolete

CRITERIA 1

The Trust Framework

Within a level & quadrant

Striking a balance between consistency and distinctiveness within a quadrant, i.e., ensuring that dimensions contribute to measuring a specific type of trust

CRITERIA 2

Producing fully distinct quadrants and dimensions, i.e., obtaining zero correlation, would indicate that the Framework’s components do not measure the same phenomena

Between quadrants

Striking a balance between consistency and distinctiveness between quadrants, i.e., ensuring that all four kinds of trust are in fact related and needed to conceptualize trust as a whole

Q U A N T I F Y I N G

T H E

F R A M E W O R K

I N T E R N A L

VA L I D AT I O N

W I T H I N

L E V E L S

Within levels: Dimensions are more closely connected within each level on average, suggesting that the Framework’s health system and vaccine levels are statistically consistent

HEA LT H S YSTE M

HEA LTH SYS TEM LE VEL

Autonomy Priority alignment Capability Fairness Confidentiality Compassion Competence Access Affordability

VA C C I N E

Benefit Relevance Safety Adequacy of info Delivery setting Agency

VA C C I N E L E V E L

Autonomy

Priority alignment

Capability

Fairness

Confidentiality

Compassion

Competence

Access

Affordability

Benefit

Relevance

Safety

0.21

0.07

0.10

0.24

0.17

0.08

0.19

0.09

0.11

0.10

0.09

0.16

0.06

0.21

0.26

0.04

0.17

0.29

0.11

0.17

0.09

0.14

0.11

0.09

0.19

0.22

0.12

0.07

0.26

-0.05

0.11

0.24

0.17

0.18

0.15

0.08

0.06

0.12

0.15

0.16

0.14

0.16

0.12

0.07

0.05

0.12

0.16

0.24

0.11

0.13

0.12

0.31

0.23

0.10

0.15

0.11

0.08

0.13 0.04 Average correlation NA within the health 0.37 0.04 0.37 NA system level

0.06

0.13

Average correlation between 0.04 0.09 0.04 0.04 vaccine0.15 dimensions and 0.06 0.08 0.17 dimensions 0.15system 0.18 0.23 0.07

0.06

0.16

0.31

0.16

0.14

0.07

0.13

0.15

0.11

0.24

0.23

0.16

0.33

0.10

0.08

0.09

0.11

0.09

0.05

0.10

Adequacy of info Delivery setting

Agency

0.04

-0.05

0.17

0.11

0.29

0.24

0.11

0.17

0.19

0.17

0.18

0.12

0.09

0.15

0.07

0.11

0.10

0.14

0.33

-0.01

0.00

0.09

0.07

0.03

0.05

0.11

0.14

0.08

0.05

0.13

0.15

0.07

0.10

-0.01

0.43

0.22

0.29

0.40

0.13

0.10

0.11

0.06

0.12

0.11

0.07

0.08

0.00

0.43

0.27

0.31

0.29

0.07

0.09

0.10

0.09

0.22

0.27

0.24 0.17

0.16 0.16 0.06

0.09

0.12

0.19

0.15

0.22

0.16

Average correlation between 0.13 0.08 0.15 vaccine dimensions and 0.09 0.15 0.18 0.15 system dimensions 0.04 0.17 0.23 0.11

0.12

0.14

0.04

0.06

0.07

0.11

0.29

0.11

0.07

0.29

0.31

0.09

0.03

0.40

0.29

0.05

0.13

0.07

0.08 Average correlation 0.28 0.41 withinNAthe vaccine level 0.24 NA

0.28

0.29

0.41

0.33

0.08

0.24

0.33

NA 48

Q U A N T I F Y I N G

T H E

F R A M E W O R K

I N T E R N A L

VA L I D AT I O N

W I T H I N

Q U A D R A N T S

Within quadrants: Using the Vaccine Trust Framework to allocate dimensions, we improve internal correlation in three out of four quadrants compared to distribution at random P RO C E D U R E

RESULTS Avg. correlation (at random)

While we have qualitative reason to group dimensions as per the Vaccine Trust Framework, we also tested internal consistency within quadrants by comparing internal quadrant correlation with the correlation coefficient had the dimensions been distributed at random.

Avg. correlation within true quadrant

Change (%)

Health system promise

0.15

0.10

-29%

Healthcare delivery

0.15

0.22

48%

Vaccine promise

0.15

0.31

111%

Vaccine delivery

0.15

0.32

122%

Health system promise is the only quadrant that doesn’t improve its internal correlation when compared with the random distribution. This is expected as the quadrant is the most abstract and difficult to measure, compounding large concepts such as fairness and autonomy. 49

Q U A N T I F Y I N G

T H E

F R A M E W O R K

I N T E R N A L

VA L I D AT I O N

W I T H I N

Q U A D R A N T S

Within quadrants: Health system promise has relatively low internal consistency due to two dimensions showing negative correlation—a challenge the Vaccine Trust Tool will address Two dimensions in Health system promise, Capability and Fairness, are negatively correlated, resulting in a reduced correlation score for this quadrant Autonomy

Priority alignment

Capability

Fairness

The Trust Tool aims to shed light on this dynamic through re-operationalization of Fairness The observed negative correlation could indicate a scenario in which some respondents view the health system as capable of treating illnesses, but also experience that they are to some extent excluded from treatment because of socio-demographic markers such as ethnicity, gender, and geography.

Autonomy

0.21

0.07

0.10

Priority alignment

0.21

0.26

0.04

Capability

0.07

0.26

-0.05

Notably, Fairness represents a complex dimension that by definition could negatively correlate with other dimensions e.g. capability as people might evaluate the health system positively for others than themselves.

Fairness

0.10

0.04

-0.05

To examine this, the Vaccine Trust Tool will revisit the operationalization of dimensions to clarify this dynamic.

CURRENT C A P A B I L I T Y

VERSIONS F A I R N E S S

Being treated differently in the health system because of …

REVISED

VERSION

F A I R N E S S

Being treated more poorly in the health system because of… PROPRIETARY AND CONFIDENTIAL | 50

Q U A N T I F Y I N G

T H E

F R A M E W O R K

I N T E R N A L

VA L I D AT I O N

B E T W E E N

Q U A D R A N T S

Between quadrants: We find that the quadrants hold unique variation when fitted in the same regression model—indicating that they are related, but also distinct RESULTS

P RO C E D U R E

System

Independent variable II

Vaccine

Independent variable I

All quadrants hold unique variation

Dependent variable

Independent variable III

To test relatedness and distinctiveness between quadrants, we fit four regression models with one quadrant as the ‘dependent’ variable and the remaining three quadrants as independent variables, assessing whether they all hold unique variation when predicting the last quadrant

All quadrants are highly significant (p-values < 0.0001) in all models. As expected, same-level quadrants produce higher coefficients, and opposite-level quadrants slightly lower coefficients. OLS Dep: Health System Promise

RESULTS*

C O E F.

Dep: Vaccine promise

C O E F.

Healthcare delivery

0.42

Health system promise

0.09

Vaccine promise

0.06

Healthcare delivery

0.14

Vaccine delivery

0.1

Vaccine delivery

0.29

Dep: Healthcare delivery

C O E F.

Dep: Vaccine delivery

C O E F.

Health system promise

0.27

Health system promise

0.24

Vaccine promise

0.05

Healthcare delivery

0.19

Vaccine delivery

0.05

Vaccine promise

0.43

*) Intercepts omitted for simplicity

Q U A N T I F Y I N G

T H E

F R A M E W O R K

I N T E R N A L

VA L I D AT I O N

M U LT I C O L L I N E A R I T Y

Our data shows multicollinearity between the four trust quadrants—but results remain statistically significant throughout when predicting vaccination behavior The Vaccine Trust Framework conceptualizes trust as four related trust types, making multicollinearity a premise for analysis The trust quadrants display low to moderate pairwise correlation—however, Variance Inflation Factors (VIF) are high (+10), suggesting that the combination of trust quadrants drive multicollinearity, and in turn, increases statistical uncertainty of results PA I RW I S E C O R R E L AT I O N * Health promise

Healthcare delivery

Vaccine promise

VIF ANALYSIS* Vaccine delivery

Health promise

0.4

0.2

0.26

Healthcare delivery

0.4

0.19

0.22

Vaccine promise

0.2

0.19

0.4

Vaccine delivery

0.26

0.22

0.4

*) Calculated based on results from Kenya on the HPV vaccine

Despite multicollinearity, we observe significant results across analyses of the four trust quadrants This indicates that although results are subject to higher statistical uncertainty, trust comes out as a strong predictor of vaccine uptake.

Q U A N T I F Y I N G

T H E

F R A M E W O R K

E X T E R N A L

VA L I D AT I O N

External validation: Compared to other widely used trust measures, the Vaccine Trust Tool turns out to add signficantly more explanation to variations in vaccine uptake Interpersonal trust measures are not helpful in explaining variation in vaccine uptake, producing statistically insignificant results. For the Vaccine Trust Tool, the correlations are stronger and highly significant indicating a strong relationship between trust and vaccine uptake.

Logistical regression correlations between vaccine uptake and different trust measures reported in log odds

C19 vaccine uptake

Interpersonal trust2

NOT

0.13

INTERPERSONAL TRUST

Interpersonal trust is a widely used and accepted measure of trust in the trust literature – often applied in studies on health-seeking behaviors and vaccine uptake. The question: “Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?” stems from the World Value Survey1.

1) World Value Survey Wave 7 (2017-2022). Documentation: The questionnaire. https://www.worldvaluessurvey.org/WVSDocumentationWV7.jsp

Childhood vaccine uptake

HPV vaccine uptake

SIGNIFICANT

0.19

-0.17

SIGNIFICANT

Vaccine Trust Tool3

2.58***

1.42**

3.68***

Correlations are controlled for respondent gender, age, educational attainment, language, religion and province. **p<0.01; ***p<0.001 2) Measured by the binary question “Generally speaking, would you say that most people can be trusted or that you must be very careful in dealing with people? ” 3) Measured as a combined measure of the four trust quadrants in the Vaccine Trust Framework: Trust in the health system promise, the healthcare delivery, the vaccine promise, and the vaccine delivery.

Methodology Sample composition Quantifying the Trust Framework

Differences in trust Trust and the effect on vaccine uptake HPV learnings

C H A P T E R

O V E R V I E W

In the following chapter, we test differences in trust across a range of socio-demographic and vaccine behaviors CHAPTER OUTLINE

S U M M A RY

Gender

Women: employment status

Caregivers of boys/girls

Education

Income: Above or below the poverty line

Survey language

Regions

Rural/urban

Vulnerability

Zero doses and vaccine rejectors

Null findings

People don’t trust uniformly across Kenya and Pakistan – women are more trusting than men, regions such as North Eastern in Kenya and Balochistan and Sindh in Pakistan are less trusting, and marginalized groups such nondominant language speakers are less trusting.

We explore differences in trust across demographic and socio-economic groups on a quadrant and dimension level 55

I N T R O D U C T I O N

T O

T H E

VA C C I N E

T R U S T

S C O R E

The reported vaccine trust score is calculated based on the HPV vaccine in Kenya, and an average trust score across vaccines in Pakistan K E N YA

PA K I S TA N

For Kenya, the reported trust scores are calculated based respondents’ answers to HPV vaccine specific questions

For Pakistan, the reported trust scores are an average calculated across respondents’ answers to COVID-19 and childhood vaccine specific questions

The Vaccine Trust Survey was designed for HPV specificities and included more questions on HPV compared to other vaccines. As a result, trust scores on HPV are more granular than other vaccines, and the results are, therefore, reported on HPV when possible.

In Pakistan, the HPV vaccine hasn’t been introduced, and its, therefore, not possible to report trust scores on it. Instead, the reported trust scores are an average across vaccines to ensure the most precise results

VACCINE OPTIONS TO REPORT TRUST SCORES ON IN THE TRUST DATA:

The trust scores can easily recalculate based on different vaccines depending on the objective of a particular analysis. KENYA

PA K I S TA N

The HPV vaccine

The COVID19 vaccine

The COVID19 vaccine Childhood vaccines Average vaccine trust across HPV, COVID19 and childhood vaccines

Childhood vaccines Average vaccine trust across COVID19 and childhood vaccines

B E H AV I O R A L

H Y P O T H E S E S

D I F F E R E N C E S

I N

T R U S T

G E N D E R

FINDING

KENYA

Women are significantly more trusting than men across all trust quandrants

Testing gender differences across quadrants

T O TA L T RU S T

Women Men

70 65

QUADRANTS

OVERALL DIFFERENCE

Health system promise

5 points (p<0.01)

Healthcare delivery

4 points (p<0.01)

Vaccine promise

Vaccine delivery

n = 2946

n = 724

Women trust the health system promise more than men

Women healthcare delivery more than men

5 points (p<0.01)

Women trust the vaccine promise on more than men

5 points (p<0.05)

Women trust the vaccine delivery more than men

INTERPRETATION

Our qualitative research did not yield any specific hypotheses in differences among men and women. However, we see that women are significantly more trusting than men across all four trust quadrants – this is not due to more healthcare interaction or exposure to the healthcare system. 57

B E H AV I O R A L

H Y P O T H E S E S

D I F F E R E N C E S

I N

T R U S T

G E N D E R

FINDING

KENYA

Across dimensions, women are more trusting than men— especially in Priority Alignment, Compassion, and Relevance

Testing gender differences by dimension

T O TA L T RU S T

Women Men

70 65

QUADRANTS

Health system promise

Healthcare delivery

Vaccine promise

Vaccine delivery

Autonomy

Priority alignment

Capability

Fairness

Confidentiality

Compassion

Competence

Access

Affordability

Benefit

Relevance

Safety

Adequacy of info

Delivery settings

Agency

n = 2946

n = 724

INTERPRETATION

Results indicate that women especially experience more Priority alignment with the HCS than men, as well as greater autonomy over health decisions Disparity in experienced Compassion of providers contributes the most to women having higher trust in healthcare delivery than men Perceived Relevance of the vaccine for the child contributes the most to women having higher trust in the vaccine promise

Results indicate that observed difference in vaccine delivery trust is driven by varying attitudes to Delivery settings. Notably, while women appear to experience less agency than men, this difference is not significant.

α = 0.05 or smaller; greyed out are not significant 58

B E H AV I O R A L

H Y P O T H E S E S

D I F F E R E N C E S

I N

T R U S T

G E N D E R

FINDING

PAK I STAN

Women in Pakistan are significantly more trusting than men across all trust quadrants—and especially in vaccine delivery

Testing gender differences across quadrants

T O TA L T RU S T

Women Men

69 64

QUADRANTS

OVERALL DIFFERENCE

Health system promise

4 points (p<0.01)

Healthcare delivery

2 points (p<0.01)

Vaccine promise

Vaccine delivery

n = 2482

n = 1252

INTERPRETATION

Women trust the health system promise more than men

Women trust healthcare delivery more than men

4 points (p<0.01)

Women trust the vaccine promise more than men

5 points (p<0.01)

Women trust the vaccine delivery more than men

Our qualitative research did not yield any specific hypotheses on differences among men and women. However, we see that women are significantly more trusting than men across all four trust quadrants – this is not due to more healthcare interaction or exposure to the healthcare system (controlled).

B E H AV I O R A L

H Y P O T H E S E S

D I F F E R E N C E S

I N

T R U S T

G E N D E R

FINDING

PAK I STAN

Across dimensions, women continue to hold higher trust scores than men— except for Affordability of healthcare.

Testing gender differences by dimension QUADRANTS

Health system promise

Healthcare delivery

T O TA L T RU S T

Women Men

69 64

Vaccine promise

Vaccine delivery

Autonomy

Priority alignment

Capability

Fairness

Confidentiality

Compassion

Competence

Access

Affordability

Benefit

Relevance

Safety

Adequacy of info

Delivery settings

Agency

n = 2482

n = 1252

INTERPRETATION

Women’s higher trust levels in this quadrant is especially driven by greater perceived Fairness of the system, but also experienced Autonomy over health decisions and the system’s Capability to treat Disparity in experienced Compassion of providers contributes the most to women having higher trust in healthcare delivery than men. Notably, women score lower on the affordability of healthcare Both perceived Benefit and Safety of the vaccine contribute the most to women having higher trust in the vaccine promise

The observed difference between men and women is driven by different attitudes towards Delivery settings for vaccinations, as well as differing experiences of having received Adequate information

α = 0.05 or smaller; greyed out are not significant 60

B E H AV I O R A L

H Y P O T H E S E S

D I F F E R E N C E S

I N

T R U S T

G E N D E R

E M P L O Y M E N T

S TAT U S

FINDING

K E N YA

Women outside the professional labor force exhibit significant lower trust in both healthcare and vaccine delivery than women who are employed.

Testing differences between employed and unemployed* women QUADRANTS

Health system promise

EMPLOYED

UNEMPLOYED

DIFFERENCE

No significant difference DISPARITY

Healthcare delivery

-2 points

Vaccine promise

No significant difference

(p<0.01)

Confidentiality

p<0.01

Access

p<0.01

Delivery setting

p<0.01

Agency

p<0.01

DISPARITY

T O TA L T RU S T

Employed women Unemployed women

70 68

Vaccine delivery INTERPRETATION

n = 2178

n = 766

-5 points (p<0.01)

The results indicate that women outside the professional labor force experience greater barriers when it comes to the process of receiving healthcare as well as the process surrounding the HPV vaccine. Most notably, women outside experience less agency, i.e., a lack of consent when it comes to the HPV vaccine (-11 points compared to employed women), as well as lower trust in the delivery settings. A possible explanation for this is that this group of women has lower trust in the schools used as vaccination sites, including the process around giving consent, i.e., Agency. *By unemployed we mean women who don’t “engage in activities for which you are paid in cash or kind”.

B E H AV I O R A L

H Y P O T H E S E S

D I F F E R E N C E S

I N

T R U S T

G E N D E R

E M P L O Y M E N T

S TAT U S

FINDING

PAK I STAN

Women outside the professional labor force exhibit lower trust in the vaccine delivery, i.e., C19 vaccine, driven by lacking trust in agency compared to formally employed women.

Testing differences between employed and unemployed* women QUADRANTS

EMPLOYED

UNEMPLOYED

DIFFERENCE

Health system promise

No significant difference

Healthcare delivery

No significant difference

Vaccine promise

No significant difference on dimension level DISPARITY

T O TA L T RU S T

Employed women Unemployed women

69 69

Vaccine delivery INTERPRETATION

n = 226

n = 2251

-5 points (p<0.01)

Agency

p<0.01

We expected that unemployed women would be less in contact with the official system, and with limited experience with the system potentially exhibit lower trust. However, there are few differences in trust among the two groups and none at a health system level as we hypothesized. Instead, we see that women out of the professional labor force exhibit lower trust in the vaccine delivery (averaged based on C19 and childhood vaccines) driven by a trust disparity in Agency where unemployed women are significantly less trusting. This suggests that unemployed women are less trusting in being asked for consent or their consent being respected by the health system. *By unemployed we mean women who don’t “engage in activities for which you are paid in cash or kind”.

D I F F E R E N C E S

I N

T R U S T

C A R E G I V E R

B O Y / G I R L

FINDING

K E N YA

Caregivers answering on behalf of boys have lower trust in the vaccine – but have similar trust in the health system as caregivers answering on behalf of girls.

Testing trust differences between caregivers answering on behalf of boys or girls

T O TA L T RU S T

Girl Boy

70 67

QUADRANTS

GIRL

BOY

DIFFERENCE

Health system promise

Not significant

Healthcare delivery

Not significant DISPARITY

Vaccine promise Vaccine delivery

76 71

70 69

Incl. Agency

Excl. Agency

n = 2068

n = 1602

6 points (p<0.01) Not significant on quadrant level

Benefit

p<0.05

Relevance

p<0.01

Safety

p<0.01

Adequacy of info

Delivery settings

p<0.01

INTERPRETATION

In Kenya, the HPV vaccine is only offered to girls and all campaigns target girls exclusively with a strong emphasis on cervical cancer prevention and is, therefore, expected that caregivers answering on behalf of girls have higher trust in the vaccine – especially the vaccine promise where caregivers answering on behalf of girls hold higher trust in Benefit and the Relevance of the HPV vaccine. 63

D I F F E R E N C E S

I N

T R U S T

C A R E G I V E R

B O Y / G I R L

FINDING

PAK I STAN

There is little difference in trust between caregivers answering on behalf of boys or girls— except for Benefit, where caregivers answering for girls are more trusting.

Testing trust differences between caregivers answering on behalf of boys or girls

Vaccine promise

2 points (p<0.01)

T O TA L T RU S T

Vaccine delivery

Not significant

n = 2068

n = 1602

Girl Boy

68 67

QUADRANTS

GIRL

BOY

DIFFERENCE

Health system promise

Not significant

Healthcare delivery

Not significant DISPARITY

Benefit

p<0.01

INTERPRETATION

With the vaccine yet to be introduced in Pakistan, we did not expect large differences across caregivers answering on behalf of girls versus boys. As expected, we see no differences on the system level and only minimal variation on the vaccine level (imputed value calculated based on COVID-19 and childhood vaccine scores). 64

D I F F E R E N C E S

I N

T R U S T

E D U C AT I O N

FINDING

K E N YA

People with no education have significant lower trust in the health system promise and the vaccine delivery than those with at least primary education.

Testing differences between respondents without education and with at least primary education QUADRANTS

NO EDU

MIN. PRIMARY

DIFFERENCE

DISPARITY

Autonomy

Health system promise

3 points (p<0.01)

Healthcare delivery

Not significant

Vaccine promise

Not significant

76 p<0.01

Fairness

DISPARITY

T O TA L T RU S T

No education Primary Secondary

67 70 69

Vaccine delivery

n = 830

n = 2840

3 points (p<0.05)

Agency

p<0.01

INTERPRETATION

We expected that people with a higher education would have higher trust as they through the education system, have contact with the official system and might also have more contact with the health system. While our hypothesis is confirmed partly in higher trust in health system promise and vaccine delivery, the differences are quite small. This is likely due to the group with minimum primary education being a very diverse group that includes many different levels of education. 65

D I F F E R E N C E S

I N

T R U S T

E D U C AT I O N

FINDING

K E N YA

Higher education doesn’t yield higher trust—instead people who have completed primary as the highest level of education is significantly more trusting than people with no or longer educations.

Testing differences between respondents without education, primary education and secondary education

T O TA L T RU S T

No education Primary Secondary

67 70 69

QUADRANTS

NO EDUCATION

SECONDARY

PRIMARY

Health system promise

+3 points (p<0.01)

0 point Not significant

Healthcare delivery

+1 points (p<0.01)

-1 point (p<0.01)

Vaccine promise

+4 points (p<0.01)

-3 point (p<0.01)

Vaccine delivery

+3 points (p<0.05)

-1 point Not significant

n = 830

n = 1286

n = 1554

INTERPRETATION

Based on our qualitative findings, we expected people with low or no education to exhibit significantly lower trust than people with longer education. Interestingly, the quantitative results suggest that people who have completed primary education are the most trusted – in both the system and the vaccine. 66

D I F F E R E N C E S

I N

T R U S T

E D U C AT I O N

FINDING

PAK I STAN

People with no education have significantly lower trust in the health system delivery and vaccine delivery – driven primarily by low trust in the dimensions access, affordability and agency

Testing differences between respondents without education and with at least primary education

T O TA L T RU S T

No education Primary Secondary

66 67 69

QUADRANTS

Health system promise

NO EDU

MIN. PRIMARY

DIFFERENCE

Not significant DISPARITY

Healthcare delivery

3 points (p<0.01)

Vaccine promise

Not significant

Access

61 p<0.01

Affordability

DISPARITY

Vaccine delivery

n = 1011

n = 2723

3 points (p<0.01)

Agency

p<0.01

INTERPRETATION

As in Kenya, we expected people with a higher education to be more trusting as they through the education system have been in contact with the official system in some capacity. Our hypothesis is confirmed partly in healthcare delivery driven by discrepancies in Access and Affordability and in vaccine delivery driven by differences in the Agency. However, the differences are quite small, likely due to the group with minimum primary education being a very diverse group that includes many different levels of education.

D I F F E R E N C E S

I N

T R U S T

E D U C AT I O N

FINDING

PAK I STAN

People with a secondary education trust the vaccine and the health system delivery significantly more than other groups with shorter or no education.

Testing differences between respondents without education, primary education and secondary education

T O TA L T RU S T

No education Primary Secondary

66 67 69

QUADRANTS

NO EDUCATION

SECONDARY

PRIMARY

Health system promise

0 points Not significant

+1 point (p<0.05)

Healthcare delivery

+2 points (p<0.01)

+2 point (p<0.01)

Vaccine promise

+1 point Not significant

+1 point (p<0.05)

Vaccine delivery

+1 point Not significant

+3 point (p<0.01)

n = 1011

n = 1252

n = 1471

INTERPRETATION

Following our expectation that higher education yields higher trust, we see that people with a minimum of secondary education are more trusting than people with shorter or no education. While the differences in trust are significant, they are still quite small indicating that while trust differs across education levels, trust in the health system or the vaccine level isn’t driven by trust.

D I F F E R E N C E S

I N

T R U S T

P O V E R T Y

L I N E

FINDING

K E N YA

People above and below the poverty line1 trust similarly. Only trust in vaccine promise is slightly but significantly higher among people below the poverty line

Testing differences between respondents above and below the poverty line1

T O TA L T RU S T

Above Below

69 69

QUADRANTS

⍏

⍖

ABOVE

BELOW

Health system promise

0 points Not significant

Healthcare delivery

0 points Not significant

62 DISPARITY

⍏

⍖

Vaccine promise

2 points (p<0.02)

Safety

p<0.05

Vaccine delivery

1 point Not significant

Agency

p<0.01

Delivery setting

p<0.05

n = 2710

n = 960

INTERPRETATION

In Kenya, we observed no stark differences between respondents above and below the poverty line, contrary to expectations. The only significant— but slight—difference is trust in the vaccine promise, which is higher for respondents below the poverty line. At the dimension level, this seems to be driven by lower trust in safety, although both groups’ low scores indicate worry about side effects and adverse events. Perceived agency in receiving vaccines differed most starkly among the two groups, with respondents above the poverty line scoring 11 points higher. 1) Poverty lines in Kenya defined as Ksh 3,947 and Ksh 7,193 per person per month for rural and urban areas respectively, as defined by the Kenya Bureau of Statistics (2021) in World Bank 2023: “Kenya Poverty and Equity Assessment 2023”;

D I F F E R E N C E S

I N

T R U S T

P O V E R T Y

L I N E

FINDING

PAK I STAN

People above and below the poverty line1 reported similar levels of trust overall, with the only significant difference being trust in healthcare delivery

Testing differences between respondents above and below the poverty line1

T O TA L T RU S T

Above Below

68 67

QUADRANTS

Health system promise

⍏

⍖

ABOVE

0 points Not significant

BELOW

Healthcare delivery

4 points (p<0.01)

Vaccine promise

0 points Not significant

Vaccine delivery

1 point Not significant

n = 2243

DISPARITY

⍏

⍖

Access

Affordability

p<0.01

n = 1491

INTERPRETATION

In Pakistan, people in both groups scored similarly across dimensions, contrary to our expectations. People above the poverty line reported higher trust in healthcare delivery, although both groups reported lowest trust in this quadrant. Looking at dimension-level scores, it appears that low trust in access and affordability are the main drivers, with the latter being very low for both groups. Even so, people below the poverty line scored 10 points lower than those above. 1) Poverty lines in Pakistan, the national poverty line is calculated as USD 3.65 per person per day (30.000 PKR per month) in World Bank (2023): “Poverty and Equity Brief Pakistan “

D I F F E R E N C E S

I N

T R U S T

L A N G U A G E

FINDING

K E N YA

People answering the survey in a nondominant language have lower trust in the health system – but are equally trusting in vaccines as people answering in a dominant language

Testing differences between respondents answering survey in English & Kiswahili or other* language

T O TA L T RU S T

English / Kiswahili Other

69 67

QUADRANTS

ENGLISH /KISWAHILI

Health system promise

Healthcare delivery

Vaccine promise Vaccine delivery

5 points (p<0.01)

OTHER

2 points (p<0.01)

2 points Not significant

0 points Not significant

n = 2854

DISPARITY

Priority alignment

73 p<0.01

Autonomy

Access

Affordability

Agency

p<0.01

n = 812

INTERPRETATION

Using language as a proxy for proximity to power at a national scale, we expected that people answering in a non-dominate language (not English or Kiswahili) to be less trusting than dominant language speakers as they are further away from power in Kenya and have a higher likelihood of feeling politically marginalized or excluded from e.g., health campaigns. As hypothesized, people answering in a non-dominant language proved to have lower trust in, especially the health system, suggesting that people further away from power are less trusting in the health system promise and healthcare delivery. *Languages included in other: Kikamba, Kikuyu, Kisii, Luhya Lou, Masssai, Nadi, Somali, Turkana & Other

D I F F E R E N C E S

I N

T R U S T

L A N G U A G E

FINDING

PAK I STAN

People answering the survey in another language than Urdu are significanly less trusting than people answering the survey in Urdu across all trust quadrants

Testing differences between respondents answering in Urdu or other language

Vaccine promise

T O TA L T RU S T

Vaccine delivery

Urdu Other

68 65

QUADRANTS

URDU

OTHER

DIFFERENCE

DISPARITY

Priority alignment

Health system promise

-2 points

Healthcare delivery

-2 points

-4 points (p<0.01)

Relevance

-3 points

Delivery settings

n = 3027

n = 706

(p<0.05)

(p<0.01)

p<0.01

Fairness

Confidentiality

Affordability

Benefit

p<0.01

INTERPRETATION

Using language as a proxy for proximity to power at a national scale, we expected that people not answering in Urdu to be less trusting than Urdu speakers as they are further away from the centers of power in Pakistan and have a higher likelihood of feeling politically marginalized or excluded from e.g., health campaigns. As hypothesized, people who didn’t answer the survey in Urdu proved to have lower trust in both the health system and vaccines, suggesting that people further away from power are less trusting. *Languages included in other: Balochi, Brahvi, Hindko, Kashmiri, Punjabi, Pushto, Sindi & Other

D I F F E R E N C E S

FINDING

The North Eastern region stands out among the regions as having lower trust – especially in the vaccine

North Eastern Nyanza Rift Valley Western

R E G I O N S

0,66

0,75 0,59

0,69 0,71 0,69

0,8

0,4

0,2

0,0

System x Process 0,66 0,63

0,59 0,64 0,59 0,60

Intervention x Promise Intervention x Process

1,0 0,64 0,60

0,8

0,4

0,2

0,0

0,73 0,69

0,72 0,70 0,71 0,70

0,6

0,4

0,76 0,74 0,78

0,69 0,47

0,0

System x Promise

1,0

0,74 0,75 0,73

0,6

0,4

73 70 66 70 57 69 70 68

Nairobi

0,75 0,72

0,6

Eastern

1,0

0,8

Coast

T R U S T

1,0

T O TA L T RU S T

Central

I N

K E N YA

0,67

0,38

Central n = 461

Eastern n = 688

North Eastern n = 119

Rift Valley n = 869

Coast n = 352

Nairobi n = 369

Nyanza n = 514

Western n = 298

Total n = 3670

INTERPRETATION

Kenya's’ Northeastern region is experiencing high levels of conflict stemming from ethnic clashes, increasing drought and food insecurity and military operations along the border to Somalia - insecurity that are impeding healthcare and fruitful vaccination campaigns, and most likely result in significant lower trust compared to other more stable and less deprived regions.

D I F F E R E N C E S

I N

T R U S T

R E G I O N S

N O R T H

E A S T E R N

R E G I O N

FINDING

K E N YA

At the health system, the North Eastern region have particularly low trust score on Autonomy, Priority alignment, and Compassion relative to other regions

Testing health system trust differences between North Eastern and rest of Kenyan regions

T O TA L T RU S T

North Eastern Rest

57 69

HEALTH SYSTEM DIMENSIONS

REST AVG

DIFFERENCE

P-VALUE

Autonomy

-15 points

<0.01

Priority alignment

-24 points

<0.01

Capability

-1 point

Not significant

Fairness

-7 point

<0.05

Confidentiality

-6 points

<0.05

Compassion

-11 points

<0.01

Competence

0 points

Not significant

Access

-4 points

Not significant

Affordability

6 points

<0.01

n = 119

n = 3670

INTERPRETATION

The North Eastern region in Kenya is poor and economically marginalized and local clans fight over the power in the region. Combined, these factors result in shifting power dynamics and experienced distance to the government power. We expected this to result in lower trust health system, and the quantitative analysis confirmed our expectations. This result suggests that distance to government power negatively impacts trust in the system. 74

D I F F E R E N C E S

I N

T R U S T

R E G I O N S

N O R T H

E A S T E R N

R E G I O N

FINDING

K E N YA

At the vaccine level, North Eastern region has significantly lower trust in the Benefit and Relevance of vaccines, as well as Adequacy of info and Delivery settings

Testing vaccine trust differences between North Eastern and rest of Kenyan regions VACCINE DIMENSIONS

Rest

57 69

REST AVG

DIFFERENCE

P-VALUE

Benefit

-50 points

<0.01

Relevance

-21 points

<0.01

Safety

-3 points

Not significant

Adequacy of info

-15 points

<0.01

Delivery setting

-38 points

<0.01

Agency

22 points

<0.01

n = 119

n = 3670

T O TA L T RU S T

North Eastern

INTERPRETATION

Following the situation in the Northern Region with high insecurity and disrupted health campaigns, we expected trust in vaccines to be lower in this region compared to other regions, and the results confirmed our expectations. People are significantly less trusting in the Benefit of the HPV vaccine and the Delivery Setting – corresponding well with the volatile security situation in the region and the disrupted and often government-led HPV campaigns. 75

D I F F E R E N C E S

I N

T R U S T

U R B A N / R U R U A L

FINDING

K E N YA

Rural areas have significantly lower trust in the health system, while urban areas have significantly lower trust in the vaccines.

Testing trust differences between rural and urban areas*

T O TA L T RU S T

Urban Rural

69 69

QUADRANTS

RURAL

URBAN

OVERALL DIFFERENCE

Health system promise

2 points (p<0.01)

Healthcare delivery

2 points (p<0.05)

Vaccine promise

3 points (p<0.01)

Vaccine delivery

3 points (p<0.01)

n = 1125

n = 2545

Rural respondents have lower trust in the health system promise

Rural respondents have lower trust in healthcare delivery

Urban respondents have lower trust in the vaccine promise

Urban respondents have lower trust in the vaccine delivery

INTERPRETATION

Following the hypothesis that distances to power both in terms of actual distance, ethical ties, and economic prosperity negatively impact trust in the health system, the results confirm our expectation. Interestingly trust in the vaccine is higher in rural areas. However, the trust differences across urban and rural areas are quite small which suggests that urban/rural isn’t a strong driver of differences in trust.

D I F F E R E N C E S

FINDING

Across regions, Balochistan and Sindh have significantly lower trust than the rest of the country, especially in the vaccine promise and delivery. T O TA L T RU S T

Azad Jammu & Kashmir Balochistan Gilgit Baltistan ICT Khyber Pakhtunkhwa Punjab Sindh

I N

T R U S T

R E G I O N S

PAK I STAN

72 62 70 71 69 69 64

1,0

1,0 0,8 0,6

0,67

0,66 0,62 0,61 0,65 0,66 0,64

0,8

0,4

0,2

0,2 System x Promise System x Process

1,0 0,8 0,6

0,59 0,57

0,64 0,63 0,64 0,62 0,60

0,63

0,6

0,4

0,0

0,79

0,0

1,0 0,8

0,75 0,75 0,73

0,69

0,62

Intervention x Promise Intervention x Process 0,79

0,66

0,76 0,75 0,71 0,72

0,67

0,6

0,4

0,2

0,0

Azad Jammu & Kashmir n = 153

Gilgit Baltistan n = 156

Khyber Pakhtunkhwa n = 417

Balochistan n = 178

ICT n = 155

Punjab n = 1697

Sindh n = 978 Total n = 3734

INTERPRETATION

In Pakistan, we expected that poorer or politically marginalized regions with less influence on national politics would be less trusting in the health system and in the vaccines introduced and rolled out by the system. As expected, Balochistan and Sindh proved to be the least trusting regions, scoring lower on the health system level and at a vaccine level compared to others. Both Sindh and Balochistan are among the poorest regions in the country and host the largest numbers of e.g., Afghan refugees and internally displaced people. In addition, both regions especially Balochistan experience political marginalization. 77

D I F F E R E N C E S

I N

T R U S T

R E G I O N S

B A L O C H I S TA N

FINDING

PAK I STAN

At a health system level, Balochistan have particular low trust in Fairness and Affordability, but are at the same time more trusting on Priority Alignment compared to other regions

Testing health system trust differences between Balochistan and rest of Pakistani regions

T O TA L T RU S T

Balochistan Rest

62 68

HEALTH SYSTEM DIMENSIONS

REST AVG

DIFFERENCE

P-VALUE

Autonomy

-2 points

Not significant

Priority alignment

5 points

<0.05

Capability

0 points

Not significant

Fairness

-19 points

<0.01

Confidentiality

-8 points

<0.01

Compassion

0 points

Not significant

Competence

0 points

Not significant

Access

-6 points

<0.01

Affordability

-10 points

<0.01

n = 178

n = 3734

INTERPRETATION

In Balochistan, we expected trust scores to be lower than the rest of the country, especially in the health system promise and healthcare delivery due to poverty, discrimination and political marginalization. As expected, Balochistan are less trusting especially in Fairness and Affordability.

D I F F E R E N C E S

I N

T R U S T

R E G I O N S

B A L O C H I S TA N

FINDING

PAK I STAN

At the vaccine level, Balochistan has lower trust in especially the Benefit and Relevance of the vaccine (avg. of childhood & C19), as well as the Delivery settings

Testing vaccine trust differences between Balochistan and rest of Pakistani regions VACCINE DIMENSIONS

Rest

62 68

REST AVG

DIFFERENCE

P-VALUE

Benefit

- 6 points

<0.01

Relevance

- 6 points

<0.01

Safety

- 3 points

<0.01

Adequacy of info

- 3 points

Not significant

Delivery setting

- 6 points

<0.01

Agency

- 2 points

Not significant

n = 178

n = 3734

T O TA L T RU S T

Balochistan

INTERPRETATION

Balochistan has one of the lowest vaccine coverage rates in Pakistan, and we expected trust at the vaccine level to be lower than other regions. We found that people in Balochistan are less trusting especially in the vaccine promise (imputed based on C19 and childhood vaccines).

D I F F E R E N C E S

I N

T R U S T

R E G I O N S

S I N D H

FINDING

PAK I STAN

At the health system level, Sindh has a signficant lower trust score on Fairness in particular, as well as Confidentiality

Testing health system trust differences between Sindh and rest of Pakistani regions

T O TA L T RU S T

Sindh Rest

64 69

HEALTH SYSTEM DIMENSIONS

REST AVG

DIFFERENCE

P-VALUE

Autonomy

-5 points

<0.01

Priority alignment

-3 points

<0.01

Capability

-2 points

<0.01

Fairness

-8 points

<0.01

Confidentiality

-6 points

<0.01

Compassion

-5 points

<0.01

Competence

0 points

Not significant

Access

-2 points

Not significant

Affordability

2 points

Not significant

n = 978

n = 3734

INTERPRETATION

The trust differences in Sindh compared to other regions are less pronounced than for Balochistan but still significant. At a health system level , Sindh is especially less trusting on Fairness which align with our expectations for a poorer and more politically marginalized region. While we see no significant differences on Affordability or Access as expected, Sindh does score lower on Confidentiality and Compassion which indicate a general lower trust in healthcare providers. 80

D I F F E R E N C E S

I N

T R U S T

R E G I O N S

S I N D H

FINDING

PAK I STAN

At the vaccine level, Sindh has lower trust all dimensions – especially on Benefit and Relevance, as well as Adequacy of info and Delivery settings compared to other regions

Testing vaccine trust differences between Sindh and rest of Pakistani regions VACCINE DIMENSIONS

Rest

64 69

REST AVG

DIFFERENCE

P-VALUE

Benefit

-8 points

<0.01

Relevance

-12 points

<0.01

Safety

-3 points

<0.01

Adequacy of info

-12 points

<0.01

Delivery setting

-7 points

<0.01

Agency

-2 points

<0.05

n = 978

n= 3734

T O TA L T RU S T

Sindh

INTERPRETATION

Following the hypothesis that distances to power both in terms of actual distance, ethical ties, and economic prosperity, negatively impact trust at the health system level, the results confirm our expectations. However, the trust differences across urban and rural areas are quite small which suggests that urban/rural isn't a strong driver of differences in trust. 81

FINDING

PAK I STAN

Rural areas trust healthcare delivery less than uban areas, while urban areas trust the vaccine prpomise less.

Testing trust differences between rural and urban areas:

T O TA L T RU S T

Sindh Rest

64 69

QUADRANTS

RURAL

URBAN

OVERALL DIFFERENCE

Health system promise

Not significant

Healthcare delivery

3 points (p<0.01)

Vaccine promise

2 points (p<0.01)

Vaccine delivery

Not significant

n = 1820

n = 1914

Rural respondents have lower trust in healthcare delivery

Urban respondents have lower trust in the vaccine promise

INTERPRETATION

Following the hypothesis that distances to power both in terms of actual distance, ethical ties and economic prosperity, negatively impacts trust at the health system level , the results confirms our expectations. However, the trust differences across urban and rural areas are quite small which suggest that urban/rural isn’t a strong driver of differences in trust. *In Pakistan, urban is defined as cities with 100.000– 1 million or inhabitants, where rural is defined as villages with 5000 inhabitants or less

D I F F E R E N C E S

I N

T R U S T

V U L N E R A B I L I T Y

FINDING

K E N YA

There are no clear differences in trust when comparing the four vulnerability segments from the Pathway vulnerability tool.

Testing differences between the four vulnerability segments*

T O TA L T RU S T

69 Less 70 More/Most 65 Least

QUADRANTS

LEAST VULNERABLE

Health system promise

- 1 points Not significant

-2 point Not significant

- 1 point Not significant

Healthcare delivery

0 points Not significant

- 5 point Not significant

5 point Not significant

Vaccine promise

2 point p<0.01

- 7 point Not significant

1 point Not significant

Vaccine delivery

3 point p<0.05

- 7 point Not significant

- 3 point Not significant

n = 1325

LESS VULNERABLE

n = 1876

MORE VULNERABLE

n = 30

MOST VULNERABLE

n = 435

INTERPRETATION

Based on our ethnographic research, we expected that the most vulnerable populations groups (more and most vulnerable) would be less trusting especially in the promise and the process of the system driven by low trust in priority alignment and fairness at the promise level and low trust in access and affordability at the process level. However, there are no clear or significant differences in trust when we compare all four vulnerability segments. This might be explained by the big variation in population size, where the least vulnerable groups are strongly overrepresented. *The vulnerability segments are based on the Pathway Vulnerability Typing Tool

D I F F E R E N C E S

I N

T R U S T

V U L N E R A B I L I T Y

FINDING

K E N YA

The most vulnerable group is significantly less trusting compared to the least vulnerable – especially in the vaccine delivery

Testing differences between the least and the most vulnerability segments*

T O TA L T RU S T

69 Less 70 More/Most 65 Least

QUADRANTS

MOST VULNERABLE

LEAST VULNERABLE

Health system promise

Healthcare delivery

Vaccine promise

Vaccine delivery

- 4 points p<0.01

0 points Not significant

- 4 point p<0.05

-7 point p<0.01

n = 1325

69 DISPARITY

Agency

p<0.01

n = 435

INTERPRETATION

As we expected, the most vulnerable population is significantly less trusting than the least vulnerable group. While the trust disparity is evident in the health system as we expected, there are no differences in trust in healthcare delivery. The most vulnerable groups are significantly less trusting in vaccines especially the vaccine delivery which is driven by a lack of trust in Agency, suggesting that the most vulnerable people fear their consent won't be heard or respected. *The vulnerability segments are based on the Pathway Vulnerability Typing Tool

D I F F E R E N C E S

I N

T R U S T

Z E R O

D O S E

R E J E C T O R S

FINDING

K E N YA

HPV vaccine rejectors have significantly lower trust in the HPV vaccine - especially in the vaccine delivery

Testing differences between respondents who accept and those who reject the HPV vaccine QUADRANTS

Rejected Accepted

65 73

✓

REJECTED1

ACCEPTED1

Health system promise

3 points Not significant

Healthcare delivery

2 points Not significant

Vaccine promise T O TA L T RU S T

Vaccine delivery

64 n = 127

5 points (p<0.01)

9 points (p<0.01)

73 n = 890

DISPARITY

✓

Benefit

p<0.01

Relevance

p<0.01

Safety

p<0.01

Adequacy of info

p<0.01

Delivery settings

p<0.01

INTERPRETATION

In Kenya, caregivers rejecting childhood vaccines have lower trust in the health system and the vaccine. This indicates that caregivers rejecting childhood vaccines, generally, are less trusting in the health system, whereas parents rejecting the HPV vaccine showed similar trust levels in the system compared to HPV vaccine acceptors. 1) “Rejected” are defined as respondents who answered ”Yes” to being offered the vaccine, but “No” to having received it. “Accepted” are defined as respondents who answered “Yes” to being offered and receiving the vaccine.

D I F F E R E N C E S

I N

T R U S T

Z E R O

D O S E

R E J E C T O R S

FINDING

K E N YA

Childhood vaccine rejectors have significantly lower trust in the health system and in the vaccine promise and delivery

Testing differences between respondents who accept and those who reject childhood vaccines 2 QUADRANTS

Healthcare delivery

Rejected Accepted

60 72

Vaccine delivery

✓

REJECTED1

Health system promise

Vaccine promise T O TA L T RU S T

61 n = 57

7 points (p<0.01)

8 points (p<0.01)

7 points (p<0.01)

14 points (p<0.01)

ACCEPTED1

DISPARITY

✓

Benefit

p<0.01

Relevance

p<0.01

Safety

p<0.01

Adequacy of info

p<0.05

n = 3595

INTERPRETATION

In Kenya, caregivers rejecting childhood vaccines have lower trust in both the health system and in the vaccine. This indicates that caregivers rejecting childhood vaccines generally are less trusting in the health system, whereas parents rejecting the HPV vaccine showed similar trust level in the system compared to HPV vaccine accepters. 1) “Rejected” are defined as respondents who answered ”Yes” to being offered the vaccine, but “No” to having received it. “Accepted” are defined as respondents who answered “Yes” to being offered and receiving the vaccine. 2) Childhood vaccines is defined as measles, hepatitis B, yellow fever, DFP or other vaccines the respondent characterizes as childhood vaccines

D I F F E R E N C E S

I N

T R U S T

Z E R O

D O S E

R E J E C T O R S

FINDING

PAK I STAN

Childhood vaccine rejectors have significantly lower trust in the vaccine – especially the vaccine promise driven by lower trust in the benefit and relevance of childhood vaccines

Testing differences between respondents who accept and those who reject childhood vaccines 2

T O TA L T RU S T

Vaccine delivery

Rejected Accepted

63 68

QUADRANTS

Health system promise

Healthcare delivery

Vaccine promise

✓

REJECTED1

67 n = 317

1 points Not significant

0 points Not significant

11 points (p<0.01)

5 points Not significant

ACCEPTED1

DISPARITY

✓

Benefit

p<0.01

Relevance

p<0.01

Safety

p<0.01

72 n = 3405

INTERPRETATION

In Pakistan, both rejectors and accepters have lower trust in the health system, while caregivers rejecting childhood vaccines also have lower trust in the vaccines—especially the vaccine’s promise. This disparity is driven by significantly lower trust in the benefit and the relevance of childhood vaccines. 1) “Rejected” are defined as respondents who answered ”Yes” to being offered the vaccine, but “No” to having received it. “Accepted” are defined as respondents who answered “Yes” to being offered and receiving the vaccine. 2) Childhood vaccines is defined as measles, hepatitis B, yellow fever, DFP or other vaccines the respondent characterizes as childhood vaccines

D I F F E R E N C E S

I N

T R U S T

N U L L

F I N D I N G S

We have explored differences in trust across multiple variables with no distinct results or clear patterns INCOME

We hypothesized that high income groups would be more trusting but found no clear evidence: • In Kenya, the trust score in Affordability increased by 4%- points from lowest income group to highest income group. • In Pakistan, low-income groups had slightly lower trust in healthcare delivery driven by discrepancies on Access and Affordability.

M A R I TA L S TAT U S

When testing the differences in trust across marital status e.g., married, divorced etc., we found no clear differences in trust levels.

AGE

We hypothesized that younger respondents either would 1) would have higher trust in the vaccine because they themselves have been close to the target group for the HPV vaccine, or 2) have lower trust in the vaccine because they are more digital and as a result potentially more exposed to misinformation. We observe no significant patterns on a health system level, and a small significant difference in the vaccine promise.

NUMBER OF CHILDREN

EXPOSURE T H E O RY

We hypothesized that families with more children had more contact with health services and would have higher trust in the health system. We found no significant relationship between number of children and trust, when we calculated the total number of sons and daughters in the household.

We hypothesized that that women are more exposed to the healthcare system and that could explain their higher trust level. trust. In Kenya, women are more exposed to hospital providers, but not community providers. In Pakistan, women are less exposed to the healthcare system than men overall, disproving the hypothesis.

Methodology Sample composition Quantifying the Trust Framework Differences in trust

Trust and the effect on vaccine uptake HPV learnings

C H A P T E R

O V E R V I E W

In the following chapter, we test the effect of trust on vaccine uptake, how the different trust types interact and how trust shape women and marginalized groups CHAPTER OUTLINE

The effect of trust on vaccine uptake

We present the key argument – that trust drives vaccine uptake across vaccines and outline the marginal effect of trust

Key trust interactions

We outline how the four trust types interact with each other

S U M M A RY

Key learnings: Trust & Gender

Key learnings: Marginalized groups

We explore how the effect of trust on vaccine uptake differs across genders

We explore how trust influences vaccine uptake for marginalized groups – approximated by region and survey language

•

Trust drives vaccine uptake – when trust is high, so is the predicted likelihood of vaccine

•

Trust drives women to action before men when women have health decision power

•

For the most marginalized, the effect of trust on vaccine uptake is just as strong as for others

T R U S T

VA C C I N E

U P TA K E

K E N YA

Across vaccines, building trust pays off Observed uptake

Level of trust

Predicted uptake had trust been high1

HPV

63%

73%

C19

62%

75%

Childhood vaccines

98%

>99%

K E N YA

The trust dividend

Predicted probability of vaccination

Plot: Total trust’s effect on predicted likelihood of vaccination for HPV, COVID19, and childhood vaccines in Kenya

*)Childhood vaccination rates are higher from the onset due to childhood vaccines' long history within maternal and pediatric healthcare and the nature of the data sample that skews towards higher childhood vaccination rates . 1) Predicted uptake with total trust at 80 on average 91

T R U S T

VA C C I N E

U P TA K E

PA K I S TA N

Across vaccines, building trust pays off Observed uptake

Level of trust

Predicted uptake had trust been high1

C19

84%

95%

Childhood vaccines

91%

97%

PAK I STA N

The trust dividend

Predicted probability of vaccination

Plot: General trust’s effect on predicted likelihood of vaccination for HPV, COVID19, and childhood vaccines in Pakistan

T R U S T

VA C C I N E

U P TA K E

K E N YA

The predicted likelihood for vaccine uptake in Kenya based on the total trust scores are as follows… PREDICTED

UPTAKE

Min., 1st quartile, median, 3rd quartile and max. value of the general trust score

1st quartile

Median

3rd quartile

Max.

HPV

22%

53%

64%

72%

83%

C19

17%

51%

64%

74%

86%

Childhood vaccines1,2

90%

98%

99%

>99%

1) Childhood vaccines is defined as measles, hepatitis B, yellow fever, DFP or other vaccines the respondent characterizes as childhood vaccines.

K E N YA

Min.

2) The high prevalence of childhood vaccine makes increased trust less impactful when driving vaccine uptake, because vaccine coverage is high from the onset (98% in Kenya and 92% in Pakistan based on our sample ). 93

T R U S T

VA C C I N E

U P TA K E

PA K I S TA N

The predicted likelihood for vaccine uptake in Kenya based on the total trust scores are as follows… PREDICTED

UPTAKE

Min., 1st quartile, median, 3rd quartile and max. value of the general trust score

1st quartile

Median

3rd quartile

Max.

C191

47%

79%

86%

91%

96%

Childhood vaccines2,3

76%

89%

92%

94%

96%

1) In Pakistan, trust played a smaller role in COVID19 vaccination that was mandated upon the population. 2) Childhood vaccines is defined as measles, hepatitis B, yellow fever, DFP or other vaccines the respondent characterizes as childhood vaccines.

PA K I S TA N

Min.

3) The high prevalence of childhood vaccine makes increased trust less impactful when driving vaccine uptake, because vaccine coverage is high from the onset (98% in Kenya and 92% in Pakistan based on our sample ). 94

T R U S T

VA C C I N E

U P TA K E

K E N YA

In Kenya, the trust quadrants’ average marginal effects on vaccine uptake are as follows… ALL M ARGINAL E FFE CT S K E N YA

SUPPRESSOR EFFECT

HPV

COVID-19

0%-pts

-19%-pts

Healthcare delivery

21%-pts

-16%-pts

3%-pts

Vaccine promise

29%-pts

60%-pts

6%-pts

Vaccine delivery

28%-pts

44%-pts

Health system promise

(Not significant)

Childhood vaccines1,2

-1%-pts

(Not significant)

2%-pts

We are observing a suppressor effect when it comes to the relationship between trust in the health system promise and C-19 vaccination in Kenya. This means the initially positive link between health system promise trust and C19 vaccination turns negative when accounting for other trust types. It implies that those trusting the health system promise also trust the vaccine, leaving the scenario of trusting the vaccine but not the health system promise to explain.

(Not significant)

The marginal effect explains how a change in the trust score from 0100 influences the likelihood of vaccination on average.3 1) Childhood vaccines is defined as measles, hepatitis B, yellow fever, DFP or other vaccines the respondent characterizes as childhood vaccines.

3) Statistically, the marginal effect represents the change in the probability of the dependent variable (vaccine uptake) due to a one-unit in the independent variable (trust) – in our case when trust goes from 0-100 95

T R U S T

VA C C I N E

U P TA K E

PA K I S TA N

In Pakistan, the trust quadrants’ average marginal effects on vaccine uptake are as follows… ALL M ARGINAL E FFE CT S PAK I STAN

SUPPRESSOR EFFECT

COVID-191

Childhood vaccines2,3

Health system promise

-8%-pts

-3%-pts (Not significant)

Healthcare delivery

11%-pts

5%-pts

Vaccine promise

12%-pts

6%-pts

Vaccine delivery

9%-pts

We are observing a suppressor effect when it comes to the relationship between trust in the health system promise and C-19 vaccination in Pakistan. This means the initially positive link between health system promise trust and C19 vaccination turns negative when accounting for other trust types. It implies that those trusting the health system promise also trust the vaccine, leaving the scenario of trusting the vaccine but not the health system promise to explain.

1%-pts (Not significant)

The marginal effect explains how a change in the trust score from 0100 influences the likelihood of vaccination on average.4 1) In Pakistan, trust played a smaller role in COVID19 vaccination that was mandated upon the population. 2) Childhood vaccines is defined as measles, hepatitis B, yellow fever, DFP or other vaccines the respondent characterizes as childhood vaccines.

4) Statistically, the marginal effect represents the change in the probability of the dependent variable (vaccine uptake) due to a one-unit in the independent variable (trust) – in our case when trust goes from 0-100 96

T R U S T

VA C C I N E

U P TA K E

K E N YA

Visually, the average marginal effects of the trust quadrants compare as follows… H P V VA C C I N E

C 1 9 VA C C I N E

C H I L D H O O D VA C C I N E S

-19%-pts

0%-pts

-1%-pts (Not significant)

(Not significant)

21%-pts

29%-pts

3%-pts

-16%-pts

60%-pts

6%-pts

2%-pts 28%-pts

Plot: The marginal effect of the four trust quadrants on HPV vaccine uptake

44%-pts

Plot: The marginal effect of the four trust quadrants on COVID-19 vaccine uptake

(Not significant)

Plot: The marginal effect of the four trust quadrants on childhood vaccine uptake 97

T R U S T

VA C C I N E

U P TA K E

PA K I S TA N

Visually, the average marginal effects of the trust quadrants compare as follows… C 1 9 VA C C I N E

C H I L D H O O D VA C C I N E S

-3%-pts

-8%-pts

(Not significant)

11%-pts

5%-pts

12%-pts

9%-pts

Plot: The marginal effect of the four trust quadrants on COVID-19 vaccine uptake

6%-pts

1%-pts (Not significant)

Plot: The marginal effect of the four trust quadrants on childhood vaccine uptake 98

T RU S T A N D T H E E F F E C T O N VA C C I N E U P TA K E

Key trust interactions

T R U S T

VA C C I N E

U P TA K E

T R U S T

I N T E R A C T I O N S

High trust in the vaccine promise helps establish resilient demand—also for the people with the lowest trust in vaccine delivery High trust in the vaccine promise can compensate for lower trust in the vaccine delivery

Plot: Predicted probability of HPV vaccination in Kenya for fixed levels of vaccine delivery trust, as vaccine promise trust increases H P V

—

K E N YA

Probability of HPV vaccination with vaccine promise at 80, and vaccine delivery at minimum

In Kenya, we observe a notable increase in HPV uptake for all levels of vaccine delivery trust as vaccine promise trust increases, including those with minimal trust in the vaccine delivery 47% probability of HPV

vaccination if trust in the vaccine promise is at 80, and trust in the vaccine delivery at minimum

100

T R U S T

VA C C I N E

U P TA K E

T R U S T

I N T E R A C T I O N S

Plot: Predicted probability of childhood vaccination in Pakistan for fixed levels of vaccine delivery trust, as vaccine promise trust increases C H I L D . V A C 1 — P A K I S T A N

98% probability of HPV

In Pakistan, high vaccine promise trust can also compensate for low trust in vaccine delivery, leading to high childhood vaccines coverage for all levels of vaccine delivery

vaccination if trust in the vaccine promise is at 80, and trust in the vaccine delivery at minimum

Probability of childhood vaccination with vaccine promise at 80, and vaccine delivery at minimum2

0.8

1) Childhood vaccines is defined as measles, hepatitis B, yellow fever, DFP or other vaccines the respondent characterizes as childhood vaccines 2) Coverage is already high in our sample (91%) producing high baseline vaccination rates

101

T R U S T

VA C C I N E

U P TA K E

T R U S T

I N T E R A C T I O N S

A lack of trust in the delivery of healthcare introduces a negative spillover to vaccine uptake—even when people do trust a specific vaccine Low trust in healthcare delivery partly outweighs the positive effect of trust in vaccine promise— making high trust in health delivery a critical lever to lift vaccine uptake In Kenya: For people with low trust in healthcare delivery1, building trust in the HPV vaccine has a smaller effect on uptake compared to the rest of the population

Plot: Predicted likelihood of HPV vaccination for lowest quartile of health delivery trust compared to rest as vaccine trust increases H P V

—

K E N YA

The difference in marginal effect of trust in the vaccine for those in lowest quartile of trust in health delivery compared to rest3

-19%-points difference in effect of trust in the HPV vaccine for those in lowest quartile of healthcare delivery trust compared to rest

Trust dividend for low trust in healthcare delivery

1) Lowest quartile of healthcare delivery 2) Childhood vaccines is defined as measles, hepatitis B, yellow fever, DFP or other vaccines the respondent characterizes as childhood vaccines 3) When trust in the vaccine increases from 0-100 and controlling for the other trust types

Trust dividend for medium to high trust in healthcare delivery

-19%-pts

Average difference in effect of increasing vaccine trust between low healthcare delivery trust1 and rest

102

T R U S T

VA C C I N E

U P TA K E

T R U S T

I N T E R A C T I O N S

A lack of trust in the delivery of healthcare introduces a negative spillover to vaccine uptake—even when people do trust a specific vaccine Low trust in healthcare delivery outweighs the positive effect of trust in vaccine promise—making high trust in health delivery a critical lever to lift vaccine uptake In Pakistan: For people with medium to high trust in healthcare delivery, building trust in childhood vaccines1 increases the already high uptake—but for those with low trust in healthcare delivery,1 we barely see an effect

Plot: Predicted likelihood of childhood vaccination for lowest quartile of health delivery trust compared to rest as vaccine trust increases C H I L D . V A C 2 — P A K I S T A N

Trust dividend for medium to high trust in healthcare delivery

No trust dividend for low trust in healthcare delivery, with slope remaining flat

-7%-points difference in effect of trust in the C19 vaccine for lowest quartile of healthcare delivery trust compared to rest

0.5

1) Childhood vaccines is defined as measles, hepatitis B, yellow fever, DFP or other vaccines the respondent characterizes as childhood vaccines 2) Lowest quartile of healthcare delivery 3) When trust in the vaccine increases from 0-100 and controlling for the other trust types

The difference in marginal effect of trust in the vaccine for those in lowest quartile of trust in health delivery compared to rest3

-7%-pts

Average difference in effect of increasing vaccine trust between low healthcare delivery trust1 and rest

103

T R U S T

VA C C I N E

U P TA K E

C O N T R O L

F I G H T

A N

E P I D E M I C

Kenya exemplifies how building trust in the vaccine delivery has a strong, positive trust dividend on uptake if combined with a moderate level of trust in the vaccine promise Plot: Predicted C19 vaccination in Kenya for fixed levels of vaccine promise trust, as vaccine delivery trust increases

K E N YA

In Kenya, trust in vaccine delivery and vaccine promise increases C19 uptake— combined, we observe that for vaccine promise, ‘a little trust goes a long way’ to obtain high coverage

70% predicted vaccination rate for median vaccine promise, and vaccine delivery at 80 58% predicted vaccination rate for lowest quartile of vaccine promise, and vaccine delivery at 80

Increasing trust in vaccine delivery increases likelihood of C19 vaccination for all levels of vaccine promise trust, but the effect is lower for those with minimal trust in the C19 vaccine promise—and not enough on its own to secure containment due to low baseline vaccination rates.

Predicted probability of C19-vaccination for different levels of vaccine promise trust, had vaccine delivery been high

Increase in uptake as vaccine promise goes from minimum to 1st quartile

19% predicted vaccination rate for minimum vaccine promise, and vaccine delivery at 80

Vaccine promise trust Predicted probability of C19-vaccination with vaccine delivery trust at 80

M I N I M U M

L O W E S T Q U A RT I L E

M E D I A N

19%

58%

70% 104

T R U S T

VA C C I N E

U P TA K E

T R U S T

I N T E R A C T I O N S

In Pakistan, combining trust in the vaccine delivery with a mandate resulted in high vaccine uptake Plot: Predicted C19 vaccination in Pakistan for fixed levels of vaccine promise trust, as vaccine delivery trust increases

92% predicted probability of C19 vaccination

87% predicted probability of C19 vaccination

PAK IS TA N

80% predicted probability of C19 vaccination 68% predicted probability of C19 vaccination

Vaccine promise trust Predicted probability of C19vaccination with vaccine delivery trust at minimum

In Pakistan, the C19 roll-out combining vaccine delivery efforts with a mandate to be vaccinated affects predicted uptake in two distinct ways… High baseline vaccination rates despite low levels of trust: For Pakistan, we observe high baseline vaccination rates for all irrespective of trust in the C19 vaccine promise and delivery, suggesting that force played a significant role in ensuring uptake Vaccine delivery trust has a strong effect for the least trusting in the C19 vaccine promise: In parallel, we observe that the effect of vaccine delivery trust is particularly strong for those with minimal trust in the vaccine promise—its benefit, efficacy and safety—most likely accelerated by the mandate

M I N I M U M

L O W E S T Q U A R T I L E

68%

80%

Vaccine promise trust Predicted probability of C19vaccination with vaccine delivery trust at 80

M I N I M U M

L O W E S T Q U A R T I L E

87%

92% 105

T RU S T A N D T H E E F F E C T O N VA C C I N E U P TA K E

Key learnings: Gender

106

T R U S T

V U L N E R A B L E

G R O U P S

W O M E N

In Kenya, any level of trust has a higher effect on vaccine uptake for women than men when women report to have health decision power—making trust a key instrument to lift women’s health Definition: Health decision power Respondents who have answered “I am in charge of child health decisions” to the question “Who in your household is most in charge of child health decisions, such as when and where to seek care when a child is sick?” are defined to hold health decision power.

Trust drives women to act before men when women have health decision power: Women will then accept a vaccine on substantially lower levels of trust in the vaccine compared to men K E N YA Plot: Predicted HPV vaccination for men and women caregivers overseeing healthcare decisions for their children in Kenya as total trust increases

70% likelihood of HPV vaccination

Women Trust score on 72

Men Trust score on 92

Women with health decision power:

72%

of women in Kenya identified themselves as main decisionmaker when it came to their child’s health (n = 2121)

A trust score difference of 20 points to achieve the same predicted vaccine uptake

1) The data sample in Kenya skews towards higher income groups. This might result in a higher share of women who report to have health decision power compared to the true population.

Women with health decision power have 70% likelihood of HPV vaccinating their adolescent with a general trust score on 72… … Men with health-decision power, however, need a general trust score on 92 to reach a 70% likelihood of HPV vaccinating their adolescents This effect persists across vaccines in Kenya, where women report to have high health decision power1 but not in Pakistan. 107

T R U S T

V U L N E R A B L E

G R O U P S

W O M E N

While trust can drive people to vaccinate, trust appears to play a smaller role for women’s vaccination decisions in Pakistan—even when women report to oversee children’s health Definition: Health decision power Respondents who have answered “I am in charge of child health decisions” to the question “Who in your household is most in charge of child health decisions, such as when and where to seek care when a child is sick?” are defined to hold health decision power. Women with health decision power:

37%

While trust is a strong lever for men’s health decision-making, its impact is less pronounced for women—likely stemming from entrenched patriarchal structures PAK I STAN Plot: Predicted childhood vaccination for men and women caregivers overseeing healthcare decisions for their children in Pakistan as total trust increases

With a predicted vaccination rate of 80% when trust is at minimum, the trust dividend for women is notably small

With a total trust score of 80, men’s predicted likelihood to vaccinate is 98%—4%-pts higher than women at 94%

of women in Pakistan identified themselves as main decisionmaker when it comes to their child’s health (n = 912)

1) The data sample in Kenya skews towards higher income groups. This might result in a higher share of women who report to have health decision power compared to the true population.

In Pakistan, the impact of trust appears to be less pronounced for women compared to Kenya. This might be a result of the Pakistani society where female health decision-makers often still rely on male relatives to, e.g., visit a health provider, and many women, therefore, de facto rely on men to execute health decisions – minimizing women’s decision power For the C19 vaccine, the effect of trust is also bigger for men in Pakistan compared to women 108

T R U S T

V U L N E R A B L E

G R O U P S

W O M E N

When women’s trust is at risk in Kenya, it is primarily driven by fear that their consent to a vaccine will be compromised—especially for low-income, unemployed women There are indications1 that the pattern of higher trust among women doesn’t hold when looking at the dimension of agency*, defined as trust in the process of informed consent

Unemployed women3 exhibit significant lower trust in vaccine delivery compared to employed women—driven by a lack of trust in the process of consent collection H P V

Subset: Below the poverty line

K E N YA

K E

HPV

C19

Trust in agency

Trust in vaccine delivery unites three dimensions:2 A D E QUAC Y

O F

D E L I V E RY

S E T T I N G

EMPLOYED

General trust

70 67

The perceived adequacy of consent collection, incl. whether people trust they will be asked for their consent and whether this decision is respected 1) Difference between genders are insignificant (p>0.05), yet breaks observed pattern of women having higher trust than men 2) For an overview of all trust dimensions, see appendix 2

/100

/10

Trust in agency is critically low for women in Kenya with monthly incomes below the poverty line. H P V EMPLOYED

Trust in the vaccine delivery

72/100

Trust in agency

63/100

I N F O

AG E N C Y

UNEMPLOYED

n = 2178

/100

n = 766

/100

n = 766

72/100 n = 570

55/100 n = 570

K E N YA UNEMPLOYED

This may speak to country-specific contexts, with Kenya generally offering more possibility for women’s decision-making, and so, the lack of agency is felt more strongly.

/100

n = 275

/100

n = 275

3) Unemployed is defined as women who ”don’t engage in activities for which they are paid in cash or kind”. The definition builds on the World Bank’s definition and is adjusted to the LMIC context with input from the project’s gender advisors. https://databank.worldbank.org/metadataglossary/world-development-indicators/series/SL.IND.EMPL.ZS 4) Poverty lines in Kenya defined as Ksh 3,947 and Ksh 7,193 per person per month for rural and urban areas respectively, as defined by the Kenya Bureau of Statistics (2021) in World Bank 2023: “Kenya Poverty and Equity Assessment 2023”; For Pakistan, the national poverty line is calculated as USD 3.65 per person per day (30.000 PKR per month) in World Bank (2023): “Poverty and Equity Brief Pakistan “ 5) With Pakistan marked by caste stratification and patriarchal structures, women’s position in the social hierarchy is more vulnerable

109

T R U S T

V U L N E R A B L E

G R O U P S

W O M E N

In Pakistan, women’s trust is more at risk when it comes to the access and affordability of healthcare in general—with the biggest trust gaps observed for low-income, employed women There are indications1 that the pattern of higher trust among women doesn’t hold when looking at the dimension of affordability* of healthcare K E

Employed women3 trust healthcare delivery less than unemployed women, with employed women living below the poverty line emerging as an especially vulnerable group lacking trust in access and affordability of healthcare C 1 9

P A K I S T A N

P K

HPV

C19

Trust in agency

Trust in healthcare delivery unites five dimensions:2 C O N F I D E N T I A L I T Y C O M P A S S I O N C O M P E T E N C E

AC C E S S The perceived ease of accessing healthcare – incl. distance, time, navigation, language barriers, and availability of medical provisions

* A F F O R DA B I L I T Y The perceived ability to get healthcare when needed without having to forego or delay treatment due to cost 1) Difference between genders are insignificant (p>0.05), yet breaks observed pattern of women having higher trust than men 2) For an overview of all trust dimensions, see appendix 2

EMPLOYED

General trust

UNEMPLOYED

68 67 /100

/100

Subset: Below the poverty line Employed women under the poverty line are less trusting in the healthcare delivery – especially in affordability and access C 1 9

P A K I S T A N

EMPLOYED

Trust in healthcare delivery

Trust in affordability

Trust in access

/100

n = 226

/100

n = 226

/100

n = 2251

46 66

/100

n = 2251

/100

n = 570

32 50

/100

n = 95

UNEMPLOYED

/100

n = 275

41 57

/100

n = 904

For a large part of Pakistani society, women aren’t expected to work when married if it is financially feasible for the family. Therefore, it can be assumed that employed women under the poverty line are working out of necessity and live under more financially precarious conditions than unemployed women under the poverty line. This aligns with the results, where employed women under the poverty line are less trusting in healthcare delivery – especially in affordability and access.

110

T RU S T A N D T H E E F F E C T O N VA C C I N E U P TA K E

Key learnings: Marginalized groups 111

T R U S T

V U L N E R A B L E

G R O U P S

M A R G I N A L I Z E D

G R O U P S

To approximate marginalization, we have selected the variables survey language and geography

Definitions Survey language

Answering the survey in a nondominant language often signifies political and financial marginalization

Geography

Regional variation covers different political, economic, and security situations across a country

North Eastern Balochistan

K E N YA

Survey language: In Kenya, speaking a non-dominant language in conversation with strangers often indicates lower socioeconomic class and marginalization Speaking English or Kiswahili signals class in Kenya. When people prefer not or or isn’t able to answer survey in English/Kiswahili, it signals both financial and political marginalization and is closely tied to people’s socioeconomic class.

Geography: The North Eastern region is one of the poorest and most politically marginalized regions in Kenya SECURITY SITUATION The Northeastern region suffers from ethnic clashes and cattle fights exacerbated by severe droughts and food insecurity. ECONOMIC SITUATION Approx. 70% of people in the Northeastern region live below the poverty line and the region suffers from infrastructure deficit, including few roads and limited access to water.

PAKI S TA N

Survey language: Preferring not to or not being able to speak Urdu often signals lower income and some degree of political marginalization in Pakistan In Pakistan, Urdu is the dominant political language. While taught in schools across the country, some population groups don’t speak or prefer to speak in their local language, which often signals distance to the political and financial elites and, thereby, potential marginalization.

Geography: Balochistan is one of the most politically and financial marginalized regions in Pakistan SECURITY SITUATION For many years, Balochistan has faced significant security issues along the Afghan border, destabilizing everyday life for the population. ECONOMIC SITUATION More than 60% of people in Balochistan live below the poverty line and Balochistan has historically housed a significant number of Afghan refugees, who live in temporary dwellings with poor infrastructure and access to healthcare

112

T R U S T

V U L N E R A B L E

G R O U P S

M A R G I N A L I Z E D

G R O U P S

In Kenya, marginalized groups – both measured by language and geography - show increased vaccine uptake when trust increases, just like the rest of the population The likelihood of nondominant language speakers1 and people from the North Eastern region to be vaccinated increases significantly and comparably to the rest of the population when general trust2 increases.

K E N YA

Survey language G E N E R A L T RU S T H P V VAC C I N E

Geography

G E N E R A L T RU S T & C OV I D 1 9 VAC C I N E *

Comparable effect of trust on vaccine uptake

Non-dominant language n = 812

Kiswahili/English n = 2854

Comparable effect of trust on vaccine uptake

North Eastern n = 119

Rest of Kenya n = 3551

There is no significant difference between the effect of trust on vaccine uptake between dominant language speakers (English/Kiswahili) and non-dominant language speakers2 or between the North Eastern region and the rest of Kenya. This suggests that the effect of trust on HPV vaccination is comparable for marginalized groups and the rest of the population in Kenya 1) Non-dominant languages in Kenya: Kikuyu, Kikamba, Luhya, Somali, Kisii, Luo, Nadi, Turkana, Maasai 2) Trust, in this case, refers to a total trust score that compounds all four trust quadrants for given vaccine

*Due to the small sample size in each region, the results are indicative of a pattern, but the conclusions are not statistically robust 113

T R U S T

V U L N E R A B L E

G R O U P S

M A R G I N A L I Z E D

G R O U P S

Similarly, marginalized groups in Pakistan– both measured by language and geography - show increased vaccine uptake when trust increases, just like the rest of the population The likelihood of nonUrdu speakers1 and people from Balochistan to be vaccinated increases significantly and comparably to the rest of the population when general trust2 increases.

PAK IS TA N

Survey language

Geography

Comparable effect of trust on vaccine uptake

G E N E R A L T RU S T C H I L DVA C .

Non-Urdu n = 706

Urdu n = 3027

Comparable effect of trust effect of trust on vaccine uptake on vaccine uptake

G E N E R A L T RU S T & G E N E R A L T RU S T & C H I L DVA C . C H I L DVA C .

Balochistan n = 178

Rest of Kenya n = 3556

There is no significant difference between the effect of trust on vaccine uptake between Urdu and non-Urdu speakers1 or between Balochistan and the rest of Pakistan. This suggests that the effect of trust on childhood vaccination is comparable for marginalized groups and the rest of the population in Pakistan 1) Non-dominant languages in Pakistan: Punjabi, Sindhi, Pushto/Pashto, Balochi, Kashmiri, Hindko, Brahvi. 2) Trust, in this case, refers to a total trust score that compounds all four trust quadrants for given vaccine

*Due to the small sample size in each region, the results are indicative of a pattern, but the conclusions are not statistically robust 114

Methodology Sample composition Quantifying the Trust Framework Differences in trust Trust and the effect on vaccine uptake

HPV learnings 115

C H A P T E R

O V E R V I E W

In the following chapter, we outline perceptions pertaining to the HPV vaccine

CHAPTER OUTLINE K E N YA

Vaccine perception

Reasons for refusing the HPV vaccine

Reasons for accepting the HPV vaccine

PAK I STAN

Intermediaries and influencers

Preliminary learnings: HPV-vaccine in Pakistan

We examine the perceived benefits of the vaccine and the social and vaccine-specific beliefs that shape vaccine acceptance and refusal, as well as trusted influencers in the context of HPV across respondents, gender, and vulnerability.

We present the preliminary learnings on the HPV vaccine in Pakistan based on the small subset of the Pakistani.

116

H P V- S P E C I F I C

L E A R N I N G S

VA C C I N E

P E R C E P T I O N

The HPV vaccine offers a dual advantage in caregivers’ minds — it protects their child’s health, but also their community as a whole 74%

K E N YA

Single choice question In general, how important do you think it is for children to get the HPV vaccine to protect your community?

5% 6%

6% 7%

Not at all important

A little important

Not at all important

5% 10% A little important

Answering for girls (n=2057)

FINDING

14% 15% 2% 3% Moderately important

Single Choice Question: How important is it for [Child] to get the HPV vaccine, also known as the ‘cervical cancer vaccine’ to protect her/his health, if available?

15%

69%

Very important

76% 60%

14% 14% Moderately important

Don’t know

1% 2% Very important

Don’t know

Answering for boys (n=1589)

The vast majority of caregivers answering on behalf of boys and girls report that the HPV vaccine is “Very important” to protect both their child’s health, but also their wider community As expected, caregivers answering on behalf of boys skew lightly towards less importance, especially when it comes to the importance of the HPV vaccine for their son’s health 117

H P V- S P E C I F I C

L E A R N I N G S

R E A S O N S

F O R

A C C E P T I N G

Perceived vaccine effectiveness — coupled with general vaccine acceptance — appears to drive uptake among caregivers whose daughters are vaccinated against HPV Single choice question: Do you agree with the following statement? 100

93%

90 80

K E N YA

94%

58%

HPV vaccine-specific beliefs: Respondents accepting the vaccine believe in the effectiveness of the HPV vaccine—with no notable difference in its protectiveness against HPV (92.7%) and cervical cancer (91.9%)

40 30 20 10 0

M O S T I M PA C T General vaccine acceptance: The most reported reason for accepting the HPV vaccine is that respondents in general believe that vaccines are important to get for their health (93.9%)

70%

70 60

93%

92%

L E A S T I M PA C T Everyone I know accepted the HPV vaccine for their children

The HPV vaccine will protect my daughter’s future

Pct. Answering Yes.

The HPV vaccine will prevent my daughter from getting cervical cancer

Somebody I trust told me it was important to vaccinate my daughter against HPV

The HPV vaccine prevents my daughter from getting HPV

I generally think vaccines are important to get

n = 890

Norm driven health-seeking behavior: While still impactful, fewer respondents report that everyone around them accepted the vaccine (57.8%)

Options: Yes, No, Don’t Know PROPRIETARY AND CONFIDENTIAL | 118 118

H P V- S P E C I F I C

L E A R N I N G S

R E A S O N S

F O R

R E F U S I N G

Caregivers who refused their daughters’ vaccination against HPV report concerns about side effects and a lack of information about the vaccine 100

Single choice question: Do you agree with the following statement?

M O S T I M PA C T

73%

Lacking information: The majority of respondents (73%) list a lack of information as a reason for not accepting the vaccine

K E N YA

50%

48% 49%

40 30

24% 24%

20%

10 0

27% 13%

11%

14%

L E A S T I M PA C T

The HPV My Somebody I am I am I am My religion Everybody I need I want to I am Most vaccine daughter worried worried I trust told against is against I know more wait until worried family is not has a very me it was vaccines vaccines refused to that the information about the about side members effective low risk of minor side important in general let their HPV on the vaccine effects or close getting effects, not to daughters vaccine vaccine has been that will friends HPV such as vaccinate get the will around affect my opposed fever or a my HPV promote longer daughter’ the swollen daughter vaccine risky s future vaccine arm against behavior HPV

Pct. Answering Yes. Options: Yes, No, Don’t Know

Side effects: Respondents also report concerns around both short-term (50%) and long-term (49%) side effects of the HPV vaccine

n = 127

Religion: A minority of respondents cite their religion as a reason for not accepting the vaccine (6%) Norm driven behavior and opinions: Both opposing opinions from family and close friends (14%) as well as seeing surroundings refuse the vaccine (11%) play a minor role in the decision-making around the HPV vaccine PROPRIETARY AND CONFIDENTIAL | 119 119

H P V- S P E C I F I C

L E A R N I N G S

I N T E R M E D I A R I E S

I N F L U E N C E R S

Healthcare providers appear to hold the potential to play a pivotal role in increasing uptake among caregivers who initially refused the HPV vaccine 100 80

89%

Single choice question: Would you accept the HPV vaccine for [daughter] if it was recommended by…

M O S T I M PA C T Healthcare providers: The vast majority of caregivers report willingness to vaccinate against HPV if explicitly recommended by a healthcare provider (89%)

K E N YA

65% 60

54%

53%

Family: 65% of caregivers who have refused the vaccine do report willingness to vaccinate if their family recommended it

37%

40 20 0

2% Your family?

Your friends?

Pct. Answering Yes. Options: Yes, No, Don’t Know

Their teacher?

Your Church?

A religious A healthcare figure (e.g., provider? a pastor, priest, sheikh, imam)?

Others?

n = 127

7/127

Caregivers who refused the vaccine will not accept the vaccine regardless of who recommended it

56/106

Caregivers are “Unlikely” to vaccinate their child (regardless of having been offered the vaccine) would accept the vaccine if a healthcare provider recommended it 120

H P V- S P E C I F I C

L E A R N I N G S

G E N D E R

There is little difference among female and male caregivers when it comes to their reasoning for accepting the HPV vaccine for their daughters Single choice question: Do you agree with the following statement?

93% 93%

100 90

92%

K E N YA

70 60 50

93% 91%

89%

94% 94%

For both women and men, general vaccine acceptance as well as believing in the effectiveness of the HPV vaccine remain the most reported reasons for having accepted the vaccine

71% 59%

M O S T I M PA C T

63% 51%

40 30 20 10 0

Everyone I know The HPV The HPV accepted the vaccine will vaccine will HPV vaccine for protect my prevent my their children daughter’s future daughter from getting cervical cancer

Women (n = 780) Pct. Answering Yes.

Options: Yes, No, Don’t Know

Men (n=110)

Somebody I trust told me it was important to vaccinate my daughter against HPV

The HPV vaccine prevents my daughter from getting HPV

I generally think vaccines are important to get

D I S PA R I T Y Light indications that social factors in the decision-making process—seeing people around you accepting the vaccine and having it recommended by a trusted individual—are slightly more prevalent for women than men (+8 %-points, +8 %-points) PROPRIETARY AND CONFIDENTIAL | 121 121

H P V- S P E C I F I C

L E A R N I N G S

G E N D E R

Similarly, men and women emphasize the same reasons for refusing the vaccine – both men and women report lacking information and concerns around side effects 100 90

Single choice question: Do you agree with the following statement?

M O S T I M PA C T 74%

K E N YA

50% 52%

50 40 30 20

24% 35% 22% 21%

10 0

The HPV vaccine is not effective

20% 17% 13% 13% 7%

49% 51% 43% 39% 27% 26% 11% 13%

Men (n=23)

13%

17%

My I am Somebody I am My religion daughter worried I trust told against is against has a very about me it was vaccines vaccines low risk of minor side important in general getting effects, not to HPV such as vaccinate fever or a my swollen daughter arm against HPV

Women (n=104)

70%

Everybody I I am I need I want to I am Most family know worried that more wait until worried members refused to the HPV information the vaccine about side or close let their vaccine will on the has been effects friends daughters promote vaccine around that will opposed get the risky longer affect my the vaccine HPV behavior daughter’ vaccine s future

For both women and men, a lack of information about the HPV vaccine is the most prevalent reason for not accepting the HPV vaccine, followed by concern of short-term and long-term side effects D I S PA R I T Y Despite the small sample, more men than women believe their daughters to have very low risk of getting HPV(+14 %-points compared to women)—while more women than men worry about long-term side effects (+12 %-points compared to men)

Pct. Answering Yes. Options: Yes, No, Don’t Know

PROPRIETARY AND CONFIDENTIAL | 122 122

H P V- S P E C I F I C

L E A R N I N G S

G E N D E R

Women report more willingness to accept the vaccine if recommended by their surroundings and network 100 90 80

K E N YA

70 60

M O S T I M PA C T 74%

67% 59%

52%

57%

38% 30%

39%

35%

Healthcare providers remain the primary influencer whose endorsement would prompt most men and women to reconsider the HPV vaccine

55%

50 40

92%

Single choice question : Would you accept the HPV vaccine for [daughter] if it was recommended by…

43%

D I S PA R I T Y

Your family? Your friends?

Women (n=104)

Their teacher?

Your Church?

Men (n=23)

Pct. Answering Yes. Options: Yes, No, Don’t Know

A religious A healthcare figure (e.g., a provider? pastor, priest, sheikh, imam)?

Others?

While we observe disparity across all actor types, results indicate that teachers are especially divisive with 59% of women reporting a willingness to accept the vaccine if recommended by teachers, compared to only 35% of men

PROPRIETARY AND CONFIDENTIAL | 123 123

H P V- S P E C I F I C

L E A R N I N G S

V U L N E R A B I L I T Y

General vaccine acceptance and trusting the effectiveness of the HPV vaccine remain the most impactful factors in accepting the vaccine across all vulnerability segments Single choice question: Do you agree with the following statement? 100

92%

95% 89%

92% 93%

92%

86%

80 67%

K E N YA

70 60 50

61%

72%

94%

89%

95% 93% 94%

General vaccine acceptance continues to be the most reported factor across vulnerability segments, alongside confidence in the effectiveness of the HPV vaccine against both HPV and cervical cancer

68%

59%

51%

40 30 20 10 0

Everyone I know The HPV accepted the vaccine will HPV vaccine for protect my their children daughter’s future

Least (n=286) Pct. Answering Yes. Options: Yes, No, Don’t Know

The HPV vaccine will prevent my daughter from getting cervical cancer

Less (n=516)

Somebody I trust told me it was important to vaccinate my daughter against HPV

MoreMost (n=87)

M O S T I M PA C T

The HPV vaccine prevents my daughter from getting HPV

I generally think vaccines are important to get

D I S PA R I T Y Light indications that more vulnerable populations are driven slightly more by general vaccine acceptance (94%) than confidence in the HPV vaccine (89% HPV; 86% cervical cancer) when comparing responses across vulnerability segments PROPRIETARY AND CONFIDENTIAL | 124 124

H P V- S P E C I F I C

L E A R N I N G S

V U L N E R A B I L I T Y

Lack of information and concerns around side effects remain the most reported reasons for refusing the vaccine across vulnerability segments 100 90

Single choice question: Do you agree with the following statement?

83% 76%

M O S T I M PA C T

75%

68%

67%

70 51%

46%

K E N YA

50 40 30 20

52% 50% 51%

23% 29% 25% 24% 25% 19%

44%

20% 25% 19% 17% 13%

10 0

3% 0%

27%

9% 10%

41%

27% 25%

13% 8%

25% 15% 11%

The HPV My I am Somebody I am My religion Everybody I am I need I want to I am Most family vaccine daughter worried I trust told against is against I know worried more wait until worried members is not has a very about me it was vaccines vaccines refused to that the information the \ about side or close \ effective low risk of minor side important in general let \ntheir HPV on the nvaccine effects \ nfriends getting effects, not to daughters vaccine \ vaccine has been nthat will opposed HPV such as vaccinate get the nwill around affect my the vaccine fever or a my HPV promote longer daugher’s swollen daughter vaccine risky future against arm behavior HPV

Least (n=59)

Less (n=59)

MoreMost (n=12)

Pct. Answering Yes.

Options: Yes, No, Don’t Know * It important to bear in mind that our sample is smaller and therefore more uncertain for the ”MostMore” vulnerable

segment

Caregivers across vulnerability segments report needing more information about the HPV vaccine as well as concerns around side effects*

D I S PA R I T Y Our sample of more or most vulnerable populations who have refused the vaccination of their daughter is small (n=12); however, in this group of people, concerns around side effects appear to play a larger role compared to lesser vulnerability segments PROPRIETARY AND CONFIDENTIAL | 125 125

H P V- S P E C I F I C

L E A R N I N G S

V U L N E R A B I L I T Y

Healthcare providers continue to be the influencer whose recommendation would make people reconsider taking the HPV vaccine – no matter the vulnerability segment 100

Single choice question: Would you accept the HPV vaccine for [daughter] if it was recommended by…

90 80

K E N YA

M O S T I M PA C T

83%

73%

70 60

100% 93%

58%

64%

58%

53%

51%

49%

48%

61%

Healthcare providers remain the most trusted influencer across vulnerability segments, despite the smaller, and more uncertain sample

55% 42%

27%

33%

D I S PA R I T Y

0% Your family?

Your friends?

Least (n=59)

Their teacher?

Less (n=59)

Your Church?

A religious figure (e.g., a pastor, priest, sheikh, imam)?

A healthcare provider?

Others?

Results suggest that the Less vulnerable are generally more willing to reconsider HPV vaccination based on recommendations from their surroundings and network

MoreMost (n=12)

Pct. Answering Yes. Options: Yes, No, Don’t Know 126

Preliminary learnings: HPV-vaccine in Pakistan 127

H P V- S P E C I F I C

L E A R N I N G S

PA K I S TA N

Prior to the introduction of the HPV vaccine, respondents across Pakistan have very low awareness about the Human Papilloma Virus, and in turn, the vaccine against it

PA K I STAN

AWA R E N E S S OF H P V

of respondents in Pakistan have heard about the Human Papilloma Virus

240 / 3734

Measured by the question: Have you heard about the infection human papillomavirus, also known as HPV?

AWA R E N E S S O F T H E H P V VA C C I N E

of respondents in Pakistan have heard about the HPV vaccine

181 / 3734

Measured by the question: Have you heard about the HPV vaccine, also known as the "cervical cancer vaccine”? 128

H P V- S P E C I F I C

L E A R N I N G S

PA K I S TA N

The majority agrees that vaccines that targets adolescent girls specifically are suspicious – indicating that the HPV vaccine will raise question in Pakistan 100

PA K I STAN

Single choice question: Please tell me if you do not at all agree, somewhat agree, or strongly agree: “Vaccines that are specifically made for adolescent girls are suspicious”

FINDING

Almost one-third of the respondents strongly agree that vaccines specifically targeting adolescent girls are suspicious

60 40 20 0

34% 20%

Not at all agree

28% 18%

Somewhat agree

Strongly agree

Don’t know

The majority (62%) agree that vaccines specifically for adolescent girls are suspicious – 28% strongly agree, and 34% somewhat agree. This suggests that the HPV vaccine will raise concerns if introduced only to girls in Pakistan.

N = 3715

PROPRIETARY AND CONFIDENTIAL | 129 129

H P V- S P E C I F I C

L E A R N I N G S

PA K I S TA N

Across regions, people are suspicious of vaccines for girls only; especially so in Baluchistan and Sindh, where three-quarters were skeptical 100% 90%

Single choice question: Please tell me if you do not at all agree, somewhat agree, or strongly agree: “Vaccines that are specifically made for adolescent girls are suspicious”

FINDING

80%

PA K I STAN

70% 60%

58%

50%

37% 37%

40% 30% 20% 10% 0%

23%

Azad Jammu & Kashmir

43% 36% 34%

26% 21% 18% 15%

21%

12% 7%

38%

31% 25%

13%

23% 20%

Not at all agree

Gilgit Baltistan

ICT

Somewhat agree

13%

6% Balochistan

35%

31% 27% 21% 21%

Khyber Pakhtunkhwa

Strongly agree

Punjab

Sindh

Don’t know

Residents in Baluchistan and Sindh are notably more suspicious of vaccines specifically targeting adolescent girls compared to the rest of Pakistan In Baluchistan, 74% agree that vaccines specifically for adolescent girls are suspicious – 37% strongly agree, and 37% somewhat agree. In Sindh, 78% agree that vaccines specifically for adolescent girls are suspicious – 43% strongly agree, and 35% somewhat agree. This suggests that the HPV vaccine will particularly raise concerns in these regions if introduced only to girls. PROPRIETARY AND CONFIDENTIAL | 130 130

H P V- S P E C I F I C

L E A R N I N G S

PA K I S TA N

People are most likely to accept the HPV vaccine if recommended by family or healthcare providers – suggesting the importance of activating health providers in a future HPV vaccine rollout 100 90

Single choice question: Would you accept the HPV vaccine if it was recommended by…

FINDING

PA K I STAN

70 60 50 40

36%

18%

27%

36% 18%

Your family?

Your friends?

Their teacher?

A religious A healthcare figure (e.g., provider pastor, (e.g., a priest, nurse, sheikh, doctor, lady imam)? health worker)?

Pct. Answering Yes. Options: Yes, No, Don’t Know

Others N = 3734

Even for influencers with the highest impact on the HPV vaccine decision – family and healthcare providers – only one-third of respondents are willing to accept the HPV vaccine if recommended 36% of respondents are likely to accept the HPV vaccine if recommended by either a family member or healthcare provider. Only 18% are willing to accept the vaccine if recommended by friends or religious figures. PROPRIETARY AND CONFIDENTIAL | 131 131

H P V- S P E C I F I C

L E A R N I N G S

PA K I S TA N

People are most likely to accept the HPV vaccine if recommended by healthcare providers but in some regions, people turn to family members and teachers – underscoring the importance of activating the broader social community in a future HPV roll out 70%

Single choice question: Would you accept the HPV vaccine if it was recommended by…

FINDING

63%

While people in many regions are most likely to accept the HPV vaccine if recommended by healthcare providers, they are outweighed by family members in multiple regions

60%

PA K I STAN

50% 40%

53%

51% 46% 40% 40% 40% 37% 36% 33%

37%

41%

39% 37% 37% 36%

38% 35% 32%

30%

28% 25% 19%

20%

24% 24% 21% 17%

16% 16% 15% 13%

10% 0%

34% 35%

29% 28%

1%2%0%1%1%1%2%

Your family?

Your friends?

Their teacher?

Azad Jammu & Kashmir

Balochistan

Gilgit Baltistan

Khyber Pakhtunkhwa

Punjab

Sindh

Pct. Answering Yes. Options: Yes, No, Don’t Know

A religious figure? A healthcare (e.g., pastor, provider? (e.g., a priest, sheikh, nurse, doctor, lady imam) health worker)

ICT

Others

In several regions, family—and in some cases, the child’s teacher’s— recommendations outweigh that of healthcare professionals, namely in Baluchistan (40% vs 38%), ICT (37% vs 34%), and Sindh (36% vs 28%). This underscores the importance of vaccine advice from from trusted community members, especially in areas where trust in healthcare providers is low. PROPRIETARY AND CONFIDENTIAL | 132 132

Original strategies for the worlds of tomorrow